History of the Game's Strength - The Era Difficulty Rating

[SABR Matt]

I thought the history buffs might find this one interesting enough that I decided to post it here...the "sabermetrics" involved here are very light mathematically, so it fits in.

This is just experimental, because to properly scale my difficulty rating, I had to arbitrarily choose the marginal value you'll see in a moment...I'm working on ways to more rigorously define it.

OK...bear with me for a moment while I explain where I got this idea.

I've been looking for the LONGEST time for a way to objectively rate how "deep" or "difficult" a league was...

I never liked James' subjective timeline adjustment...it seemed WAY too simple. But how do you go about seeing how skilled the players within a league are as a group?

The idea came to me through a discussion I had with Randy Fiato (TKD) about what defines "bad baseball". It is intuitively obvious that when two bad teams face each other, the games will be sloppy more frequently..mistakes will be made in all aspects of the game. Pitching mistakes...hitting mistakes...fielding mistakes...baserunning blunders.

What will this look like statistically though? A classic idea proposed by sabermetricians in the 70s was to rate players based on standard deviations from the mean...it has been observed many times that the standard deviation of batting average has been fluctuating through time but trending down...(there's a famous paper on the disappearance of the .400 hitter that discusses this...the author's name escapes me for some reason).

Batting average is not however explanatory enough...what we want to know is...does the standard deviation of run scoring per side per game change with time the way it does for batting average? Are we cycling closer and closer to the mean as time advances?

A quick survey using retrosheet.org's game logs reveals that in fact standard deviation is changing with time...but perhaps not the way you might think. It became immediately apparent that the standard deviation of run scoring on a per game basis was directly dependent on the league average run scoring rate. In fact, an r^2 of 0.9301 exists between those two variables...low scoring leagues have small standard deviations...high scoring leagues have larger standard deviations.

Does this mean that high scoring leagues are "weaker"...less deep with talent? Of course not. It's hard to argue that the deadball era was a better level of play than today's game even with expansion considernig the player pool has expanded to include approximately 50 times more potential baseball players than it did back then, minor league scouting and development didn't exist in the deadball era, and the equipment and field conditions were often horrendous, making for sloppy games far more frequently than in today's major leagues.

This dependence on run scoring environment is not however the only problem with using standard deviation to rate the difficulty of a league or the players within the league. There is a fundamental logical flaw. The use of standard normal z scores presumes that the league and/or player distribution was normal...neither is the case.

The player distribution is pyramidal...the top 1% of the humans who play baseball make the major leagues (liberally...it might be closer to .001%)...if we could rate every baseballer from tee-ball to Japan to MLB to High School...the distribution of skill might be normal. Meanwhile, the distribution of runs scored per side per game in a league is the summation of a series of one-game match-ups...each match-up behaving according to the laws of probability as governed by the intrinsic strengths of both combatants...the result of that process is a non-normal significantly skewed distribution...high extreme values will have an exaggeratedly large Z-score...shutouts are a sign of bad play too but their is a lower bound to how "bad" you can be in the non-scoring direction.

Given this lower bound...and the resulting tendency for variations in ability to manifest themselves in the rightward biasing direction (large numbers of high scoring games relative to the mean run scoring environment)...we fall back on MEASURING the skew of the league's RS distribution to get an idea about how erratic/weak that league was.

The positives...Skew is not dependent on the run scoring environemtn...it is never affected by the mean of a probability distribution. Skew uni-directional...meaning the lower bound shouldn't interfere with an accurate measurement of positive skew (skew is defined to be positive when the longer tail of a distribution points to the right on a number line). Skewness also does not presume a distribution is normal. It describes how non-normal a distribution is.

Logically...skew tells you how frequently extremes occur...more extremes mean more variation in intrinsic team strengths...and therefore...a weaker league.

If the run scoring distribution were normal (had no skew) this would mean that there was ZERO variation in player ability across the league...this would be the "ideal" league...but we know this to be humanly impossible to achieve...nonetheless...it serves to demonstrate that more skew is a larger deviation from the ideal league.

Skewness of a distribution is easily measured:

SUM(x - u)^3
--------------------
(n - 1) * s^3

Where x is the observed game/side runs scored, u is the league average runs scored per side per game, n is the number of game/sides within the league and s is the standard deviation of the distribution.

Placing the s term in the expression automatically scales the skew value so that higher scoring leagues, which will naturally have a wider range of run scoring outcomes do not appear to have higher skew.

When I plotted skew of the run scoring distribution against time, wat I found was a somewhat messy but nonetheless encouraging trend toward gradually decreasing skew with time. There was a lot of noise in the plot...probably because skew is heavily impacted by large outliers, so extreme games might have had a disproportionately large pull on skew...it therefore was necessary to smooth skew values.

I chose to use a normally weighted 7-year running mean of skew values for each league (normally weighted implies a larger emphasis on the center year...think of the shape of the bell curve) to smooth out the fluctuations...

It makes sense to smooth the data because although players change from season to season...the overall strength of the league cannot possibly fluctuate by overly large amounts...there are hundreds of players in any given league...turnover from year to year is no larger than 5-10% so we should expect league strengths to change gradually except in extreme circumstances like during WWII.

I'm considering alternatives to this normally weight running mean idea...I may for instance measure the skewness of a longer period of years than one...perhaps skew is more persistant if you incluide more than one year of data...either way...the smoothed values were eye popping and aligned very well with my expectations for where baseball was weak and where it was strong.

But this doesn't end the problem.

Assuming Smoothed skew is an appropriate measure of league strength, we need to put it in a form that allows strong leagues to score higher than weak leagues...and it would in fact be ideal if we got the scores to range from 0 to 1 so that they could be used multiplicatively...(for instance...if we rate 1872 as a 0.5 league...we would cut player wins in half in 1872 to get an idea of how many wins they'd be worth in a strong league)

We can make use of the exponential function here...it makes sense to use the exponential given that major league baseball represents the top of the baseball pyramid and the drop in skew value from typical leagues to great ones is likely to be large.

It also gives us the right range if used properly. Skewness can theoretically range from 0 to infinity in this case (it can't range negatively because of the lower bound at zero)...if we take a skewness of zero...e^0 = 1...if we take a skewness value approaching infinity e^large = large...ah but if we make that e^-skew...-0 is still zero, but -large implies 1/(e^large) which asymptotically approaches zero.

One more step though...no baseball league...no matter how great...will ever have a skew of zero. Here's the nasty part where I have to arbitrarily pick a marginal skew value. This was just me visually examining the graph of smoothed skew with time and seeing what the skew appeared to be approaching (the overall curved trend appears to be leveling off slowly but surely.

I chose a value orf 0.8 as the minimum skew...though I experimented with other values.

This was applied by simply subtracting 0.8 from each skew value obtained by the smoothing process before converting them with the exponential decay function.

The end result is quite interest to me...

Here are the top 20 most difficult leagues by this method:
Year Lg Strength
1984 AL 0.968
1985 AL 0.967
1997 AL 0.947
1995 AL 0.946
1996 AL 0.943
1998 AL 0.942
1986 AL 0.941
1983 AL 0.941
1983 NL 0.932
1933 AL 0.928
1934 AL 0.928
1999 AL 0.925
1994 AL 0.925
1982 NL 0.923
1937 AL 0.919
1935 AL 0.913
1938 AL 0.912
1936 AL 0.909
1987 AL 0.907
1962 AL 0.906

And the 20 weakest leagues
1910 NL 0.691
1909 AL 0.690
1944 NL 0.688
1902 NL 0.687
1901 NL 0.683
1885 NL 0.682
1905 AL 0.679
1911 NL 0.675
1881 NL 0.666
1875 NA 0.665
1906 AL 0.663
1908 AL 0.654
1907 AL 0.651
1874 NA 0.637
1873 NA 0.614
1884 NL 0.612
1882 NL 0.589
1872 NA 0.578
1883 NL 0.560
1871 NA 0.528

The early deadball era looks to me to have been very weak competitively...though obviously not as bad as the old National Association...which plays like a modern AA or A league.

Thoughts from the peanut gallery?

[Sultan_1895-1948]

Any way you could put this in simpler terms? I have no idea what you're tying to say. By "stronger" league, do you mean harder to excel offensively in? If thats the case, how can you have 5 seasons from 1995-present in the "hardest" list when this happened:

AVG MLB TEAM

YEAR - HR - ERA

2005 – 167 -- 4.29
2004 – 182 -- 4.47
2003 – 174 -- 4.41
2002 – 169 -- 4.28
2001 – 182 -- 4.41
2000 – 190 -- 4.77
1999 – 184 -- 4.71
1998 – 169 -- 4.46
1997 – 166 -- 4.39
1996 – 177 -- 4.61
1995 – 146 -- 4.45

[leecemark]

--What would make you assume it means hardest to excell offensively? What league difficulty measures is how good the league is and as a result how hard it is to separate from the pack in that league.

[SABR Matt]

This isn't a measure of difficulty to hit...to measure that one needs only the league's mean run scoring tendency.

This is a measure of the overall depth of the player pool and therefore the difficulty in being significantly better than average.

Both for hitters and for pitchers/fielders.

It can be measured using only the Run Scored Distribution because both the offense and the defense must contribute if a team is to score a large number of runs in a game...you have to both do your own hitting...and have the defense you face suck.

Primarily, the reason league difficulty expresses itself (at least...it appears to express itself) in a lack of extreme games is that when talent is spread more evenly...it's harder to bash a team's head in.

[Joltin' Joe]

I'm surprised that '43 & '45 does not appear in the top 20 weakest list.

[Sultan_1895-1948]

--What would make you assume it means hardest to excell offensively?

I didn't know, thats why I was asking. I'm not a big saber guy. In fact halfway through his original post I started seeing spots and had to get up for another beer :D

What league difficulty measures is how good the league is and as a result how hard it is to separate from the pack in that league.


Ok, that makes sense. If the game is easier for everyone offensively, then it would be harder to stand out from the pack. Makes sense that certain players would try to do something to stand out as well.

[SABR Matt]

It's not so much that the game is "easier for everyone" offensively...it's more than everyone is so good offensively that the league context is being dragged upward...hence making it harder to be better than the standard...but yeah...you have the idea. It also does start to explain some of the desperation by today's players toward getting an edge...

Pitchers and hitters alike.

I'm sorry if my opening post was dense..I wasn't goinfg for dense...just trying to fully explain where I got the idea for skewness research.

[SABR Matt]

As for '43 and '45...fear not...the war did have a very noticeable upward pull on both AL and NL skews...the difference was that the AL started out a much stronger league than the NL so the deleterious effects of the player loss was not as sharply felt there...and '43 and '45 in the NL are both in the top 40 weakest leagues...the curve is pretty cool actually...a perfectly timed spike in skew (drop in difficulty) right during the war.

[Sultan_1895-1948]

It's not so much that the game is "easier for everyone" offensively...it's more than everyone is so good offensively that the league context is being dragged upward...hence making it harder to be better than the standard...but yeah...you have the idea. It also does start to explain some of the desperation by today's players toward getting an edge...

Pitchers and hitters alike.

I'm sorry if my opening post was dense..I wasn't goinfg for dense...just trying to fully explain where I got the idea for skewness research.

It wasn't dense, just over my head apparently. I was done at the r^2y or whatever you put. Its all good though.

I guess its a matter of opinion as the why its harder to stand out in todays game. I would say its like restricter plates in nascar. Everybody has a strong engine (body) and MLB has given them other small advantages that add up to create more offense then we've ever seen before. The top 5% talent can no longer rise further above because its easier for the middle 40% to put up bigger numbers. Its all opinion apparently though.

[SABR Matt]

well you're not disagreeing with me...

"Everyone has strong engines (bodies)..."

That's precisely it...the atheletes are all better...batters, fielders, and pitchers are all better today than they were in 1920. The increase in offense is mostly created by the ball type and the fact that we've just had a period where the hitters were better than the pitchers (it goes in cycles...hitters dominated the 1880s and 1890s, pitchers owned the deadball era, hitters dominated in the 20s and 30s...pitchers dominated fomr the mid fifties to the early 90s...the hitters are back...that advantage is starting to wain thuogh...pitching and defense are starting to balance now...not to mention the smaller parks of today...

And BTW...it's not really true that this is the most offense we've ever had...there was more offense in 1930 and in 1894 than there is today.

[Sultan_1895-1948]

And BTW...it's not really true that this is the most offense we've ever had...there was more offense in 1930 and in 1894 than there is today.

Yeah, these parks are rediculous and so are all the other factors.

What are you basing that comment on? , runs/game?


The average team hit 98 HR in '30, and the average ERA was 4.80

The average team hit 182 HR in 2004, average ERA was 4.47

[SABR Matt]

Offense is offense...doesn't matter in this context how we came by it...I'm basing the comment on runs/game. runs/game in the modern game are higher than normal, but these years don't match the 20s/30s spike for offensive profficiency.

[Ubiquitous]

Yes bill has a timeline adjustment that is simple but it was also for a simple rating system. In the same book he has something like a ten or twelve fator timeline system. One that includes quality of play and the conditions of the playing field.

Also can you show us bottom 10 or 20 of only the 20th century? Perhaps a ranking of all the 20th century seasons. Thanks.

[SABR Matt]

Sure.

Let me do something to avoid people getting too obsessed with exact ranks of modern seasons...

Let's focus on the last 100 years of the FL, AL and NL only...that's 202 leagues from 1905 to 2004

I'm going to give the leagues letter grades in order to break the sample up into "similar" groups. I don't want people going "how could you have year X in the NL 20 ranks below year Y in the AL!" when the numerical rank is very similar so I'll give you groupings...they will be ranked in reverse order of difficulty though for anyone curious about exact ordinal rank.

Higher Difficulty - higher Letter Grade
Year Lg Gr EDR
1984 AL A+ 0.968
1985 AL A+ 0.967
1997 AL A+ 0.947
1995 AL A+ 0.946
1996 AL A+ 0.943
1998 AL A+ 0.942
1986 AL A 0.941
1983 AL A 0.941
1983 NL A 0.932
1933 AL A 0.928
1934 AL A 0.928
1999 AL A 0.925
1994 AL A 0.925
1982 NL A 0.923
1937 AL A 0.919
1935 AL A 0.913
1938 AL A 0.912
1936 AL A 0.909
1987 AL A 0.907
1962 AL A- 0.906
2000 AL A- 0.904
1932 AL A- 0.902
1961 AL A- 0.901
1984 NL A- 0.900
1982 AL A- 0.899
1963 AL B+ 0.897
2004 NL B+ 0.897
2001 AL B+ 0.891
1960 AL B+ 0.887
1993 AL B+ 0.884
2002 AL B+ 0.882
2003 AL B+ 0.880
1939 AL B+ 0.880
1998 NL B+ 0.879
2004 AL B+ 0.878
1997 NL B+ 0.877
1981 NL B 0.875
1964 AL B 0.874
1988 AL B 0.874
2003 NL B 0.872
1941 AL B 0.870
1940 AL B 0.869
1959 AL B 0.868
1999 NL B 0.867
1961 NL B 0.866
2000 NL B 0.864
1926 AL B 0.863
1962 NL B 0.863
2001 NL B 0.862
1931 AL B 0.860
2002 NL B 0.859
1981 AL B 0.858
1960 NL B 0.858
1996 NL B 0.856
1989 AL B 0.855
1942 AL B 0.854
1963 NL B 0.853
1985 NL B 0.852
1975 AL B 0.852
1927 AL B- 0.851
1992 AL B- 0.850
1964 NL B- 0.848
1974 AL B- 0.846
1965 AL B- 0.844
1959 NL B- 0.843
1990 AL B- 0.842
1958 AL B- 0.841
1991 AL B- 0.841
1928 NL B- 0.840
1980 AL B- 0.839
1976 AL C+ 0.839
1965 NL C+ 0.838
1977 AL C+ 0.838
1995 NL C+ 0.837
1978 AL C+ 0.836
1992 NL C+ 0.836
1979 AL C+ 0.835
1993 NL C+ 0.833
1925 AL C+ 0.832
1994 NL C+ 0.831
1970 NL C+ 0.831
1973 AL C+ 0.830
1948 NL C+ 0.829
1971 AL C+ 0.829
1986 NL C+ 0.829
1943 AL C 0.829
1916 NL C 0.828
1927 NL C 0.827
1971 NL C 0.826
1929 NL C 0.825
1947 NL C 0.825
1914 AL C 0.825
1915 NL C 0.825
1958 NL C 0.823
1966 AL C 0.823
1972 AL C 0.823
1928 AL C 0.821
1991 NL C 0.821
1930 AL C 0.819
1914 NL C 0.819
1970 AL C 0.818
1980 NL C 0.817
1944 AL C 0.816
1949 NL C 0.816
1946 AL C 0.815
1913 AL C 0.815
1966 NL C 0.814
1945 AL C 0.814
1947 AL C 0.813
1955 NL C 0.812
1956 NL C 0.810
1975 NL C 0.809
1972 NL C 0.808
1969 NL C 0.808
1951 NL C 0.807
1987 NL C 0.807
1940 NL C 0.807
1967 AL C- 0.807
1950 NL C- 0.807
1957 NL C- 0.806
1948 AL C- 0.806
1915 FL C- 0.806
1914 FL C- 0.806
1976 NL C- 0.805
1929 AL C- 0.805
1952 NL C- 0.805
1951 AL C- 0.803
1969 AL C- 0.803
1954 NL C- 0.803
1941 NL C- 0.802
1939 NL C- 0.801
1953 NL C- 0.800
1921 NL D+ 0.798
1974 NL D+ 0.798
1920 AL D+ 0.798
1968 AL D+ 0.797
1930 NL D+ 0.797
1973 NL D+ 0.795
1957 AL D+ 0.795
1917 NL D+ 0.795
1949 AL D+ 0.795
1952 AL D+ 0.791
1950 AL D+ 0.791
1921 AL D 0.787
1922 NL D 0.787
1977 NL D 0.787
1990 NL D 0.786
1926 NL D 0.784
1920 NL D 0.783
1988 NL D 0.782
1912 AL D 0.782
1967 NL D 0.781
1968 NL D 0.780
1938 NL D 0.780
1919 AL D 0.780
1946 NL D 0.779
1913 NL D 0.779
1915 AL D 0.778
1979 NL D 0.777
1924 AL D 0.775
1934 NL D 0.773
1931 NL D 0.773
1923 NL D 0.773
1978 NL D 0.772
1933 NL D 0.772
1932 NL D 0.770
1989 NL D- 0.768
1919 NL D- 0.762
1924 NL D- 0.762
1942 NL D- 0.761
1911 AL D- 0.761
1935 NL D- 0.761
1925 NL D- 0.760
1918 NL D- 0.759
1953 AL D- 0.757
1922 AL D- 0.756
1937 NL D- 0.755
1936 NL D- 0.746
1956 AL D- 0.746
1918 AL D- 0.738
1923 AL D- 0.736
1910 AL D- 0.731
1908 NL F 0.726
1954 AL F 0.725
1909 NL F 0.721
1945 NL F 0.721
1916 AL F 0.720
1907 NL F 0.718
1906 NL F 0.718
1955 AL F 0.715
1912 NL F 0.714
1905 NL F 0.714
1917 AL F 0.705
1943 NL F 0.704
1910 NL F 0.691
1909 AL F 0.690
1944 NL F 0.688
1905 AL F 0.679
1911 NL F 0.675
1906 AL F 0.663
1908 AL F 0.654
1907 AL F 0.651

That's from 1905-2004...thoughts?

[Ubiquitous]

I see the Cubs greatest run is during a time in the NL that you have them ranked as an F.
Is that because the league was so bad or because the Cubs were so great? In otherwords how much effect did the Cubs have on that rating? Is it possible that the team assembled so many of the better players of that era on one team that it made the whole league lower skilled when in reality it was merely disbursed unevenly?

[BillyF29]

The way I've always adjusted is very non-scientific, but is:

=1.05-((R27-ERA)/R27)

[Ubiquitous]

also can you put an attachment on your post so that one can download that list. I would like to be able to seperate out the leagues and put them in a timeline order. thanks

[538280]

I'm sure you put a lot of reasearch into this, and it looks like it's returned reasonable results, but there is one major thing I must point out. I realize you have the 1989 NL rated at a D-, and the 1989 AL rated at a B. Does that really make any sense? Why is it in a modern era that the AL should be so far ahead of the NL? That doesn't seem to make much sense to me at all. Could you perhaps explain to me why it comes out that way, or any reasonable explanation for that? To me, that could be a major flaw in this system. I can't imagine league quality between leaues was even that far, or if it was not in the modern era.

[SABR Matt]

Well...now I see the flaw in using letter grades...LOL

At least the way I've scaled them. The difference between a D- and a B isn't that huge...modern baseball from about 1950 on is very homogeneous...the spread of league difficulties is very small. I perhaps shuold have rated all of the teams in baseball history when I came up with those quick reference grades...and I definitely should have used a mean/standard deviation method instead of just breaking it up into proportions the way I did.

There was a significant difference between the AL and the NL in the late 80s though...quick...top of your head...name five great players who were at their best from 1985-1990...no cheating and looking guys up...

I did that little exercise when I was looking at the data to see if it made sense.

I obviously am not saying this is going to be perfect...now would I claim this is the end of my research...I just found it worthy of posting here because I believe I'm "on the right track"...I do believe skewness of the run scoring distribution is going to be the key to seeing league difficulty.

I have more things I want to try to smooth the data more scientifically than I did here.

I'll convery the answers I have to this point into a text file and attach them so you can recreate the nifty graph I have of difficulty by league and year.

[SABR Matt]

The way I've always adjusted is very non-scientific, but is:

=1.05-((R27-ERA)/R27)

Interesting...you tried to capture the ERROR rate...the rate at which runs score on sloppy play...as your method for era adjusting...clever idea...not bad as a quick thumbnail.

[SABR Matt]

BTW...in answer to the question about those great Cubs teams...one team's success shouldn't theoretically have an extreme impact on the entire league's run scoring distribution...nonetheless...I think it's a little of both...I think there weren't very many great players in the NL in the deadball era...but the ones there were all managed to land on a few teams...creating some really AWFUL clubs...and some good ones. I think that SHOULD have a significant negative impact on how we view those '00s Cubs...yes they had good players...but we can't be as confident that they really were GREAT as we could be if they were doing their winning in a much tougher era.

Think about it this way...the 2001 Mariners won 116 games in a league that was MUCH more demanding...who would you rather have...the '06 Cubs who won their games against horrible foes...of the '01 Mariners who contended with a couple of bad teams but largely well distributed talent?

[SABR Matt]

Attachment...

This also includes the statistical documentation you need to see how I got weighted (by league games) mean and standard deviation to calculate new letter grades...they're slightly less arbitrary and make a little more sense now...sorry about that...

If you guys want I can screencap my graph from Excel and post it here as a jpeg.

[SABR Matt]

I trust you know how to get it from text file to something you can play with in Excel, Ubiquitus...it's pretty straightforward. If anyone else wants to attempt it but doesn't know how to go from text file to excel file...let me know and I'll explain.

[Ubiquitous]

Unfortunately excel is on the fritz but after jumping through some hoops I got it to work for quattro.

Anyway I don't know if I can buy the results or not.

From 1947 on the quality of NL play is supposedly pretty mediocre, then smack dab in the middle of expansion it gets better, slide back down during the pitching era, then the latin explosion happens and the quality slides back even further and doesn't get good again until the 1980's.

Over in the American League after expansion the league stays mediocre to bad then again right around expansion gets good, falls back to mediocre for the pitching era then falls back to mediocre during the latin expansion and again doesn't get good again until the 80's.

[SABR Matt]

I quickly calculated a second EDR where instead of taking skew values for individual leagues and doing a normally weighted mean, I actually took the skew of all of the data in that seven year period.

To avoid the changing means presenting a bias, I used the (x-u) values previously calculated for one year samples...so as to make each season independent of its' run scoring mean...the result was nearly identical to the first effort except that rapid drops and increases were flattened out a little more and large gaps between leagues seemed to shrink some, particularly in the modern era.

The 1989 example brought up here for instance formerly had EDRs of .764 for the NL and .855 for the AL...the second EDR was .811 for the NL, .862 for the AL...still a noteworthy difference...though now instead of being a D+ vs a B-...it's a C vs. a B-.

[SABR Matt]

Unfortunately excel is on the fritz but after jumping through some hoops I got it to work for quattro.

Anyway I don't know if I can buy the results or not.

From 1947 on the quality of NL play is supposedly pretty mediocre, then smack dab in the middle of expansion it gets better, slide back down during the pitching era, then the latin explosion happens and the quality slides back even further and doesn't get good again until the 1980's.

Over in the American League after expansion the league stays mediocre to bad then again right around expansion gets good, falls back to mediocre for the pitching era then falls back to mediocre during the latin expansion and again doesn't get good again until the 80's.

Actually...the AL's peak in performance happens right BEFORE most of the expansion (1962 was a minor expansion in either league...the big expansions didn't happen until 1969 and 1977...both of which BTW are signularly responsible for the drop in quality of play in both leagues IMHO

The other approach I've seen to test quality of play has been W% mobility analysis...how doable is it for teams to find players and go from bad to good in short periods...there are times during the depressed 30s/40s/50s ESPECIALLY in the NL...where W% mobility is terribly low...where a couple of good teams dominated for long stretches and not many of the others had a chance to compete except for during the war where talent was essentially randomly scrambled.

The National League became seriously stagnant in the middle of the 20th century. Skewness values are astoundingly consistant...WWII aside...from 1927 to 1962 and begin slowly climbing during rapid expansion...(reflecting a drop in the quality of play)...you say "quality slides during the pitcher's era"...I say "quality slides during the EXPANSION era...and doesn't rebound until the leagues adjust to the expansion in 1981.

[Ubiquitous]

Bill James view:
The “Index of Competitive Balance,’ which is a new measurement introduced here, is composed of two elements. Those two elements are:

1. The standard deviation of winning percentages for teams in each single season during the decade, averaged.
2. The standard deviation of winning percentages among franchises for the decade as a whole.

The first of these measures the extent to which the best teams in any season are able to dominate the weakest teams. The second measures the extent to which the same teams win season after season throughout the decade.
If baseball was perfectly competitive – that is, if every team was exactly as good as every other team, and the only differences between them were in luck- then the first measure above would be .039, and the second would be .014.
The actual figures for the 1870’s were .170 and .081; I’d have to spend about three more paragraphs to fully explain the parameters used to derive these numbers, and I’m going to skip that, because it’s boring. These two figures (in each decade) are then added together, and the sum is divided by .053, which is what the sum would be in a perfectly competitive environment. This figure is then divided into 100 to produce the index of competitive balance. In other words, if the sum of these two standard deviations was .106, that would be 2.00 times what it would be in a perfectly competitive environment, which would produce an index of 50%. A perfectly competitive index is therefore 100%.
You may not have understood all of that, and you don’t need to. The essential point is that the greater the difference is between the best teams and the worst, the lower the index of competitive balance. The 1870s are the least competitive decade in baseball history, with an index of 21%.

Decade Index of Competitive Balance
1870’s 21%
1880’s 24%
1890’s 27%
1900’s 30%
1910’s 36%
1920’s 34%
1930’s 41%
1940’s 34%
1950’s 34%
1960’s 40%
1970’s 45%
1980’s 56%
1990’s 57%


Notes: (One thing) that made the races more competitive (in the 60’s) was expansion…….because a twelve-team league is inherently more difficult to dominate than an eight-team circuit.
…….The 1980’s, the first full decade of free agency, were by far the most competitive years in baseball history up to that point, and also the decade in which the small-city markets enjoyed their most success ever……In the early 1990’s this continued to be true; baseball was highly competitive, and not at all dominated by Big Market teams. But as the decade has moved on, competitive balance has begun to fray. The standard deviation of winning percentage, which was .054 in 1990 (one of the lowest figures in baseball history) jumped to .081 in 1998, the highest figure since 1977.

I have a theory that the quality of play in major league baseball, over time, could be tracked by what we could call “Peripheral Quality Indicia” - PQI, for short. Hitting by pitchers is a peripheral quality indicator; the higher the quality of play, in my opinion, the less the pitchers will hit. I have a list of about a dozen of these:

1. Hitting by pitchers
2. The average distance of the players, I age, from 27.
3. The percentage of players who are less than six feet tall or more than 6’3”
4. Fielding Percentage and Passed Balls
5. Double Plays
6. Usage of pitchers at other positions
7. The percentage of fielding plays made by the pitchers.
8. The percentage of games which are blowouts
9. The average attendance and seating capacity of the game location.
10. The condition of the field.
11. The use of players in specialized roles.
12. The average distance of teams from .500.
13. The percentage of games which go nine innings.
14. The standard deviation of offensive effectiveness.
15. The standard of record-keeping.
16. The percentage of managers who have 20 years or more experience in the game.

Ok, more than a dozen. Anyway, let’s array teams in ways which we all agree represent top to bottom:

1. Major League Baseball.
2. Minor League Baseball.
3. College baseball.
4. High School baseball.
5. Ten-year-old kids playing baseball.
6. Seven-year old kids attempting to play baseball.

If you studied that list, you would find that all of these things increased or decreased predictably as the quality of competition improved. The eigth indicator, for example, is the number of blowouts. My seven-year old son (Reuben) is on a team that lost one game 26-3, and won the next game 31-0. In high school blowouts are still common, but there are more games which aren’t blowouts. In college ball you get a few 18-0 games – more than you get in the minors or the majors. If you hear that a game has been decided 41-2, don’t you tend to assume that that was probably a low-level competition?
Batting stats and pitching stats do not indicate the quality of play, merely which part of that struggle is dominant at the moment. But fielding stats are somewhat inevitably tied to the level of competition, I ways which are reflected in the ratio between double plays and errors. In Reuben’s games, most balls in play result in errors, while I have seen only one double play all year. In my thirteen-year-old son’s games, there are still about five times as many errors as double plays. In college ball there are still more errors than double plays, but it is closer, while in the majors there are more double plays than errors.
In Reuben’s league, the average distance from age 27 is abot 20 years; in high school, about ten years, in college, about seven years, in the majors, probably three years.
In Reuben’s league, the games are attended by a handful of people; in high school, by a few dozen; in college, by hundreds; in the minors, by thousands; in the majors, by tens of thousands.
In Reuben’s league, pitchers make far more fielding plays than players at any other position. In high school, they still make as many as at any other position; in college, fewer, but still some, while in the majors the pitchers make only one or two fielding plays per game.
When kids start playing baseball the pitchers are the best hitters, in high school, the pitcher is still very often the cleanup hitter, but as they climb the ladder the pitchers hit less and less.
In Reuben’s league there are no statistics at all. In high school baseball there are sketchy statistics kept by some teams. In college ball there are statistics, but not lots of them, while for the major leagues there are nitwits like me who grind them out by the ton.
If you worked at it hard enough, you could make up a set of standards to “score” each of these things, which would track the increases in the quality of competition as Reuben moves to the major leagues, although frankly how you score the quality of the grounds keeping, I don’t want to know.
Anyway, my point is that if track major league baseball from 1876 to the present, all of these indicia, without exception, have advanced steadily. As late as the 1920’s, there were still major league managers who had little experience with the game. I know that many people passionately disagree with me when I argue that the quality of play in the majors has continued to increase, but even since 1950, all or virtually all these indicators would suggest that the quality of major league play has improved steadily.
The best-hitting pitchers of the 21st century don’t hit anything like what Bob Lemon hit, or Spahn, or Newcombe, or the other good-hitting pitchers of that era.
Success in the majors by very young players has become significantly less common (although success by old players has probably become more common).
In the 1950 major league pitchers averaged about 240 assists per team; in 2001, in a longer season, the average will be less than 200.
In 1950 there were about 1.2 double plays for each error. In 2001 the ratio will be at least 1.3 to one.
Player/managers, who were the youngest and least experienced managers, have become extinct.
The stadiums and crowds are bigger, the statistics are better, the grounds keeping standards are far higher. The teams are closer to .500. I haven’t studied it, bt I would bet there are fewer blowouts, fewer lop-sided games.
During World War II, when we could all agree that the quality of major league play dropped, these indicators reflect the drop. World War II brought into the game more players who were remote in age from 27 – more teenagers, and more old men. The double play to errors ratio, 0.86 to 1 in 1941 and advancing almost every year, dropped slightly during the war years.
When there is an expansion, these indicators reflect the drop in the quality of play. Expansion brings into the league younger players, and keeps in the league older players. Expansion pushes the standard deviation of winning percentage up and the fielding percentage down.
And yet, over time, these effects are not large enough to keep the PQI from moving higher. Is that proof that the quality of play is getting better? Perhaps it isn’t. But that is what I believe, and this is one of the reasons I believe it.

[SABR Matt]

I was also add that one of the major reasons we had our best league play in the early 80s was the onset of and correct application of FREE AGENCY...free agency seriously altered the playing field...created for teams a chance to get talent they never would have had in the dead-roster era (baseball prior to 1976) where players didn't move around much and where talent collected it had a tendency to stagnate.

[SABR Matt]

One note of caution Ubiquitus...you're comparing apples to oranges.

I'm talking about league talent DEPTH...you're talking about competitive balance...they're somewhat related but NOT the same. You can have a well balanced league where all eight teams within SUCK...it will look well balalnced...but the run scoring distribution will have large skew because from game to game, each team will be making mistakes and run scoring will happen in bunches and often.

[Ubiquitous]

I did something like this awhile back, it was just a quick and dirty query in which I measured the STD of batting averages and the league averages.

Basically what I found was a lot of the same stuff you found with your test. Deadball era does the worst, pre WWII does very well, integrations years not so hot, then the late 90's and into the 2000's having high quality.

It appears if one wants to assume that a low spread and high average as an indicator of quality that white baseball had probably hit its peak just before WWII. That the infrastucture of baseball for whites was setup well enough that they were employing the best that the white could offer.
Once integration happened in full force league quality went down. Why? Personally I think it is because it turned out that the lower tiered whites were not as good as the upper tiered blacks but they were not pushed out because the blacks were given a limited role. Meaning middle and lowered tiered blacks were still not allowed to play. Until around 1996 were the spread stays consistently small and the Average stays consistently high.

Now does that mean I personally believe the AL was in such upheaval and radically redefining itself for over 50 years? No I don't I think the pitchers and DH have some effect on the league average which would definitely effect the rankings. When I look at the NL I will know more.

I personally don't believe that the years immediately before integration and WWII as baseball's highest quality. I believe its a deception based on not letting a good chunk of talent play the game. In otherwords they created and artificial ceiling while at the same time shoring up the floor.

[Ubiquitous]

Actually...the AL's peak in performance happens right BEFORE most of the expansion (1962 was a minor expansion in either league...the big expansions didn't happen until 1969 and 1977...both of which BTW are signularly responsible for the drop in quality of play in both leagues IMHO

The other approach I've seen to test quality of play has been W% mobility analysis...how doable is it for teams to find players and go from bad to good in short periods...there are times during the depressed 30s/40s/50s ESPECIALLY in the NL...where W% mobility is terribly low...where a couple of good teams dominated for long stretches and not many of the others had a chance to compete except for during the war where talent was essentially randomly scrambled.

The National League became seriously stagnant in the middle of the 20th century. Skewness values are astoundingly consistant...WWII aside...from 1927 to 1962 and begin slowly climbing during rapid expansion...(reflecting a drop in the quality of play)...you say "quality slides during the pitcher's era"...I say "quality slides during the EXPANSION era...and doesn't rebound until the leagues adjust to the expansion in 1981.


The AL in 1961 expanded by 25%, the 69 expansion was a growth of 20%. The 1977 by 17%. All and all each expansion was an addition of two teams. How can an expansion of two teams be minor in 1961 but be major in 1969?

But nobody how we want to label the expansion both leagues improved during the first expansion and declined during the next two. Then in the 90's they improved again during expansion. It doesn't appear to me that your data supports expansion as a reason for the slide.

[SABR Matt]

Let me do something here...

I'm now using EDR2...isntead of EDR1 because I think it's logically a little more consistant to take the skew of a range of years rather than the average of a range of skews.

What I've done is to calculate DELTA...the difference between a year's EDR2 and year before it.

I want to see in which years league difficulty increased by the largest amounts from previous years and see if those times make sense...same for rapid decreases.

Top Increases:

1887 NL: probably has something to do with rules changes making for better games.
1878 NL: Dropped from attempts to field 8 teams to fielding only 6...major rules changes and longer schedules begin.
1959 AL: Negro League Baseball collapses...black wave rushes through major league baseball at every level throughout the 50s...improvements effect league depth from mid 50s to 1962.
1886 NL: See 1887 NL
1891 NL: AA players flockign to NL in droves as AA's demise becomes apparent.
1983 NL: Combination of Latin Expansion and Free Agency
1908 AL: Deadball era weakest in modern times...improvements begin as Ty Cobb and other talented players finally begin to emerge and enrich the talent pool
1981 NL: Free Agency and Latin craze (Fernando!!)
1927 AL: AL begins heading into first golden age of depth...most of the great players you can think of from the 20s played in the AL...the NL by comparison was pathetically weak.
1916 NL: Federal League players return to the majors
1994 AL: league begins recovering from 1993 expansion blip...steroids fill clubhouses

These are all making sense to me...

Let's try drop-offs:

1942 AL: WWII
1952 AL: NL's smaller parks and better scouting prevail...NL fully integrates LONG before AL does...this is why AL lags behind NL throughout the 50s
1883 AA: AA was never a strong league, but 1882 appears to be the fluke here...probably caused by lack of preceding years in the skew averaging
1947 AL: Talent shift to the NL starts right around here...
1920 AL: Hmm...this one I'm not sure about...
1986 NL: still a strong league in 1986...just coming down from a peak in '84/'85
1898 NL: A lot of teams beginning to go bankrupt...scouting of new players all but STOPPED in this period.
1951 NL: one year blip...not sure what caused this one...NL returned to previous standards shortly hereafter.

I could go on...

EDR2 looks a little more consistant than EDR1 did...but the overall impression I get is that expansion killed off the positive effects of the latin wave until free agency commensed and the latin wave really took off in the 80s...and the peak in performance just prior to expansion in the late 50s was probably due to Negro League talent.

[Sultan_1895-1948]

To me, just because its harder to stand out in a league, doesn't necessarily mean that the league is better in quality. It means that there is only so high the top 5% can go, and era differences allow the middle guys to get closer to them, making it appear more talented overall. Lower guys become middle, middle become upper, and top 5% remain, but can't dominate as much.

[SABR Matt]

You're comments were of course based on working with batting averages and the like, but you're comments about improper use of the negro leaguers does have merit...I think the reason the AL and NL both rapidly improved in the late 50s/early 60s but the AL was WAY behind the NL in mid 50s was that Negro Leaguers had more of a chance to play in the 50s NL than they did in the 50s AL...and that full integration really didn't even begin until just prior to expansion.

Notice though that rapid improvement STOPS in 1962 in both leagues and through the 60s there is retrogression...which goes on right through about 1979...expansion did weaken the game...but it rapidly recovered when FA and Latino baseball really started taking off...I think I'm on at least generally the right track here...The data appears to make sense from where I'm sitting...obviously...that's up for debate.

[SABR Matt]

Sultan...

think about it this way...if you're an elite player facing a league where there's a large difference between your performance and the second tier guys...don't you think it will be easier for you to do your job than if you're an elite player facing competition filled with a bunch of other players who are close behind you?

[SABR Matt]

Ah...I see what's going on...I'm looking at EDR2...you're still looking at EDR1...EDR2 paints a slightly different picture of the 60s...I think a more accurate one.

Here's that data for the NL:
Year Lg EDR1 LgG EDR2 DELTA
1955 NL 0.812 1232 0.802 0.005
1956 NL 0.810 1242 0.807 0.005
1957 NL 0.806 1238 0.812 0.005
1958 NL 0.823 1232 0.839 0.027
1959 NL 0.843 1240 0.840 0.001
1960 NL 0.858 1238 0.838 -0.002
1961 NL 0.866 1238 0.861 0.023
1962 NL 0.863 1624 0.852 -0.009
1963 NL 0.853 1622 0.864 0.011
1964 NL 0.848 1624 0.839 -0.025
1965 NL 0.838 1626 0.823 -0.016
1966 NL 0.814 1618 0.812 -0.011
1967 NL 0.781 1620 0.826 0.014
1968 NL 0.780 1626 0.817 -0.009
1969 NL 0.808 1946 0.818 0.001
1970 NL 0.831 1942 0.798 -0.020
1971 NL 0.826 1944 0.807 0.009
1972 NL 0.808 1860 0.818 0.011
1973 NL 0.795 1942 0.820 0.002
1974 NL 0.798 1944 0.797 -0.023
1975 NL 0.809 1942 0.798 0.002
1976 NL 0.805 1944 0.777 -0.022
1977 NL 0.787 1944 0.784 0.008
1978 NL 0.772 1942 0.796 0.012
1979 NL 0.777 1942 0.806 0.010
1980 NL 0.817 1946 0.820 0.014
1981 NL 0.875 1288 0.860 0.040
1982 NL 0.923 1944 0.838 -0.022
1983 NL 0.932 1948 0.883 0.045
1984 NL 0.900 1942 0.871 -0.012
1985 NL 0.852 1942 0.859 -0.012

There are a couple of blips out of place in the DELTA pattern...but you have to look at the general flow of the data because statistical data is never going to be perfect.

I see lots of minuses in the DELTA field through the early expansion era...then it fluctuates for a bit...then takes off after the third expansion.

That makes sense to me.

Particuarly...notice when the increases begin...1977...the first full year of free agency.

[SABR Matt]

Here's the AL over the same span...
Year Lg EDR1 LgG EDR2 DELTA
1955 AL 0.715 1236 0.749 -0.012
1956 AL 0.746 1236 0.755 0.006
1957 AL 0.795 1232 0.777 0.021
1958 AL 0.841 1238 0.798 0.022
1959 AL 0.868 1236 0.860 0.062
1960 AL 0.887 1234 0.880 0.019
1961 AL 0.901 1622 0.899 0.020
1962 AL 0.906 1618 0.883 -0.016
1963 AL 0.897 1616 0.883 0.000
1964 AL 0.874 1628 0.872 -0.011
1965 AL 0.844 1620 0.852 -0.020
1966 AL 0.823 1612 0.837 -0.015
1967 AL 0.807 1620 0.815 -0.022
1968 AL 0.797 1624 0.822 0.007
1969 AL 0.803 1946 0.813 -0.009
1970 AL 0.818 1946 0.814 0.000
1971 AL 0.829 1932 0.816 0.003
1972 AL 0.823 1858 0.845 0.028
1973 AL 0.830 1944 0.829 -0.016
1974 AL 0.846 1946 0.851 0.022
1975 AL 0.852 1926 0.835 -0.015
1976 AL 0.839 1934 0.842 0.007
1977 AL 0.838 2262 0.845 0.004
1978 AL 0.836 2262 0.842 -0.004
1979 AL 0.835 2256 0.840 -0.001
1980 AL 0.839 2264 0.869 0.029
1981 AL 0.858 1500 0.880 0.011
1982 AL 0.899 2270 0.905 0.025
1983 AL 0.941 2270 0.925 0.020
1984 AL 0.968 2268 0.937 0.012
1985 AL 0.967 2264 0.933 -0.005

The AL is even more pronounced and makes things even clearer.

1977 there was no expansion in the NL...in the AL there was...the end result...Free Agency and the expansion cancel each other out and the league holds steady. Meanwhile in the 60s the AL drops from a peak of .899 jsut prior to expansion to .813 after the '69 expansion.

[Ubiquitous]

You say it stops at 1962 but why would it continue to increase during expansion and only decline after it?

[Sultan_1895-1948]

Sultan...

think about it this way...if you're an elite player facing a league where there's a large difference between your performance and the second tier guys...don't you think it will be easier for you to do your job than if you're an elite player facing competition filled with a bunch of other players who are close behind you?

No, because the point is that players "talent" hasn't gotten better. Era factors have allowed EVERYONE to become good if not great, which closes the gap toward the top 5%.

Knowing that average players can become great because of the style of play, will push the top 5% harder (or make them cheat to gain an edge), but the my point is that there's a ceiling to the top 5%. Steroids have allowed some of these guys to poke their head through the ceiling (pun intended, ya know big craniums, ok you get it), such as 66,70, 73, but overall there's a max.

[SABR Matt]

Ubiquitus...your question confuses me...

It doesn't "continue to increase throughout expansion"...difficulty rating drops throughout the sixties after the peak in 1961 (except the blip rise in 1963 in the NL...again....there are going to be blips...but it's the overall picture)...and levels off after the 1969 expansion.

There are several factors that cause it to level off rather than continue to drop (Latin Players, increased roles for Negro Players, improvements in training, medical treatments for injuries and the rules change that lowered the height of the mound leap to mind as contributors to the stable 70s).

There is an assumption I think that the expansion shuold IMMEDIATELY and FINALLY alter the talent balance...that it should be one thing before each expansion...and another reduced thing thereafter...I don't think it quite works that way...each expansion brings with it two "bad teams"...those two teams don't really begin to suck talent out of the rest of the league for a few years...their crappiness is isolated...when the league levels out to accept the new teams...that's when the full effects of expansion are felt.

[SABR Matt]

To Sultan

I'll grant that there is possibly a ceiling to human achievement...but you're thikning only in terms of offense...league talent depth runs on both sides of the ball.

Part of the reason the 1990s register is highly deep is that although offense increased...it did not go through the roof like it did in previous eras...the batters improved...SO DID THE PITCHERS...and in fact...the in play hit rate did not climb that much from lesser offensive leagues of the 70s and 80s...which says the fielding must also have improved or at least stayed the same.

Compare that to the late 1960s where we had a run of great pitchers...but the hitters did NOT respond in kind...there was a large imbalance.

[Ubiquitous]

The AL expands in 1961, that year EDR2 is at its highest peak until the 80's. The next year it slides slightly but again is at its highest until the 80's. It holds it level in 1963 then starts it slides as the pitching era takes over.

The NL expands in 1962, in 1961 its at it highest until the 80's, it slips during 1962 but again it is at its highest until the 90's. Then in 1963 it surpasses even the high mark of 1961 before starting its slide into the pitchers era.

[Sultan_1895-1948]

To Sultan

I'll grant that there is possibly a ceiling to human achievement...but you're thikning only in terms of offense...league talent depth runs on both sides of the ball.

Part of the reason the 1990s register is highly deep is that although offense increased...it did not go through the roof like it did in previous eras...the batters improved...SO DID THE PITCHERS...and in fact...the in play hit rate did not climb that much from lesser offensive leagues of the 70s and 80s...which says the fielding must also have improved or at least stayed the same.

Compare that to the late 1960s where we had a run of great pitchers...but the hitters did NOT respond in kind...there was a large imbalance.

You say that pitching has improved, but using the naked eye along with baseball knowledge, its clearly watered down. How many guys need to be seasoned down in the minors, but clubs have too much invested in them to keep them there. How many old timers are just hangin' around because they are a bargain, and they're only asked to go 5 innings, or get 2 outs in relief. Why do you think pitchng has increased in this era? Because strikeouts are up? That shows a hitters approach.

Defense have improved because of many factors as well. Less ground to cover, smooth infields and outfields, bigger gloves, scouting charts, etc. What causes this to be rendered slightly more meaningless, is that today's style of play is slug for the fences, strikeout or nothing type of baseball.

Hey Sabermatt, do you have stats that show the number of fly ball outs recorded for each year?

[SABR Matt]

The AL expands in 1961, that year EDR2 is at its highest peak until the 80's. The next year it slides slightly but again is at its highest until the 80's. It holds it level in 1963 then starts it slides as the pitching era takes over.

The NL expands in 1962, in 1961 its at it highest until the 80's, it slips during 1962 but again it is at its highest until the 90's. Then in 1963 it surpasses even the high mark of 1961 before starting its slide into the pitchers era.

As I said Ubiquitus...it takes a few years for the effects of an expansion to be felt fully IMHO...what you basically have when you expand a league is the other teams the way they were before minus a few lesser players...and then the exapnsion teams...which are basically dogs.

It's not until the Expansion teams start pulling on the rest of the league seriously that they can really have an impact on the run scoring distribution or on the overall average level of competition faced by any one player.

[Ubiquitous]

How come that doesn't really happen at any other expansion?

One observatiomn I take away from this is that it looks to me that there never were enough white players to stock 16 teams and have good overall leagues. For instances in the 30's AL league play was high but it was at the expense of the NL. The NL during that stretch was not good.

[SABR Matt]

You say that pitching has improved, but using the naked eye along with baseball knowledge, its clearly watered down. How many guys need to be seasoned down in the minors, but clubs have too much invested in them to keep them there. How many old timers are just hangin' around because they are a bargain, and they're only asked to go 5 innings, or get 2 outs in relief. Why do you think pitchng has increased in this era? Because strikeouts are up? That shows a hitters approach.

Defense have improved because of many factors as well. Less ground to cover, smooth infields and outfields, bigger gloves, scouting charts, etc. What causes this to be rendered slightly more meaningless, is that today's style of play is slug for the fences, strikeout or nothing type of baseball.

Hey Sabermatt, do you have stats that show the number of fly ball outs recorded for each year?

I wouldn't go so far as to say the entire style of play in the 90s/00s has been "slug for the fences or nothing"...walks are also up...as hits aren't "down"...it's basically the 80s contact game PLUS more walks and power...

I have...somewhere in my huge stack of data...groundball/flyball percentages through time...and the flyball rate is in fact up in the liveball era...in fact it goes up in every liveball era because managers and players aren't stupid...they sense the ball is getting easier to hit farther/faster...they're going to try to elevate it.

So yes...fielding is being aided by less ground to cover (although the new parks added since Safeco field have been largely pitcher's parks so that trend is starting to reverse), more flyballs, and better equipment and scouting...but that's all part of the same package...the end result is that the quality of play has improved all around in the 90s...perhaps I should be careful to avoid using the word "talent" because that implies that these changes are entirely in the hands of the players...clearly...they aren't...but when you're trying to see how any player would do in a neutral environment...you have to factor out things outside his control...including the league difficulty...and a league difficulty can be raised by either improving player depth...or by improving playing conditions.

As for your comments on pitching...I'll say that the main thing that's really had a positive impact on pitching is the era of reliever specialization. The strain is coming off the starters and pitchers are being used in roles that they can maximize their abilities (lefties facing lefties...pitchers with unique deliveries being used to throw off batters in critical situations...power pitchers facing contact hitters and control artists facing pull hitters...etc)...it's easy to get bogged down in the offensive statistics and blame the pitchers...but I submit to you that the primary factor in raising offense in the recent seasons has been the live ball and the small parks...NOT the pitching.

[SABR Matt]

How come that doesn't really happen at any other expansion?

One observatiomn I take away from this is that it looks to me that there never were enough white players to stock 16 teams and have good overall leagues. For instances in the 30's AL league play was high but it was at the expense of the NL. The NL during that stretch was not good.

Answer to question first:

Actually it's happening as we speak...the 1998 expansion came at a time when baseball was again very strong...since '98 the AL weakened significantly and the NL stopped improving (problems here...interleague play...the crossing of league barriers more than ever...I expect to see the leagues become increasingly homogeneous with time...and it appears they are in fact doing that...so an expansion in any league will impact them both)...

Of course...it's not perfect...but I think it's a pretty darned good first step.

And I think we didn't see a gradual drop in the 70s because there were other factors obscurring that negative influence...the Latin surge (which was MUCH larger than the Negro surge in terms of new talent)...free agency...Tommy John surgery...and...dare I say it...sabermetrics in their fledgling forms aiding in better scouting and development...

The final graph of depth rating versus time is the result of adding and subtracting many different possible pulls on the skill of the league.

To your observation now...I FULLY agree. As I said earlier...think of ten great players from the 30s...I'll be shocked if the first ten you think of include more than one or two NL stars...the NL in the 30s was HORRIBLE...the era difficulty rating is absolutely correct in seeing a very large difference between the two leagues there...I was glad to see it sniff that out.

[Sultan_1895-1948]

but I submit to you that the primary factor in raising offense in the recent seasons has been the live ball and the small parks...NOT the pitching.

wow, I agree with a "saber" guy on something. Just playin' ;)

We can add to that : smaller strike zone, lighter and harder bats, can't come inside anymore, body armor, etc.

In the face of all this, its truly amazing that any pitchers succeed. Ironically, most of them "do" succeed, because of the overall hitters approach. Its become like a certain type of martial art (the one Segal apparently knows), where you use the others persons strength/momentum against him in a fight. This has to be the pitching approach nowdays. You're not gonna succeed by just by having a blazing fastball.

[SABR Matt]

All of that stuff...the body armor...the strike zone...that's all a part of why the game is deeper now than ever...or clsoe anyway...that came about because players and coaches got smarter and smarter...the pitchers got smarter too...they learned to counter all of that...

[Sultan_1895-1948]

All of that stuff...the body armor...the strike zone...that's all a part of why the game is deeper now than ever...or clsoe anyway...that came about because players and coaches got smarter and smarter...the pitchers got smarter too...they learned to counter all of that...

Its certainly why the game is "easier" than ever for average players. How many guys hit 25-30 HR per season. I bet its gone way up from the mediocre players now being able to reach that total.

IMO, they came about because offense = money. Baseball shows no interest in correcting these issues because all they see is $$$$$$$$$$$$ They're appealing to the lowest intellectual form of baseball fan. Not a good move if you've got the best interest of the national pastime in mind eh.

[SABR Matt]

See...I don't see it as mediocre players having an easier time clubbing 25 HRs...I see it as there being fewer truly mediocre players...I see it as baseball teams knowing they need to sock the ball to win and therefore not accepting poor offensive players as long term solutions...and of course...these players have advantages other players did not...it's easier to learn what they're doing wrong and adjust than it was in 1960...so I guess that aspect is helping some.

The best interest of the game IS for it to make lots and lots of money...the richer the game...the better the product on the field assuming the money is spent well...of course not everyone spends their money well...I do think sasbermetrics will help with that...owners will not waste money on bad players...they'll spend their money plumbing the international player pool for the good ones.

[Brian McKenna]

I don't have the exact numbers today but they are approximate.

1950s
16 major league teams, over 800 minor league teams

2000s
30 major league teams, 150 minor league teams

obviously there are fewer supplying more today. benefit: batters

[SABR Matt]

I wasn't aware there were 800 minor league teams in the 50s...I suspect however that the vast majority of those teams NEVER produced a major league player...nor came anywhere close to doing so.

Serious farm systems were more limited in the 50s than they are today...the whole method for developing players is FAR more advanced now than it was.

[Ubiquitous]

I don't have the exact numbers today but they are approximate.

1950s
16 major league teams, over 800 minor league teams

2000s
30 major league teams, 150 minor league teams

obviously there are fewer supplying more today. benefit: batters
There are more then 150 minor league teams in America. Just like in the 50's there are a lot of independent leagues out there.
Plus you have to factor in the explosion of colleges and junior colleges that didn't exist in those days or had nowhere near the level of organization that they do now.

There are over 200 teams in the minor league organization. That organization had 448 teams (an all time high) in 1949 but it quickly fell to 132 teams by 1963. As of 2004 it has 242 teams.

that of course does not include the indy leagus, of which there is at least 70 more teams.

[Ubiquitous]

Also during the golden era of minor league baseball Major league teams were reducing their farm system. In 1946 62% of all minor league teams were affiliated to a major league team. By 1951 it was down to 46%. They went from having 280 affiliates to 172 teams. So in actuality there are more affiliates today then in the golden age of minor league baseball.

In 1950 232 minor league teams that were not affiliated with a major league team. Close to the highest amount in 40 years. These teams were very close to semipro team.

Finally the height of the minors was just after WWII, the 50's was a time of decline for the minors. In 1950 there were 446 teams by 1960 there were 152 teams. I, 1950 210 were affiliated, by 1960 only 126. 33 million people went to see minor league games in 1950, only 10.6 million went to see them in 1960.

[SABR Matt]

I think it is universally understood that minor league player development is better now than at any time in major league history.

[csh19792001]

As I said Ubiquitus...it takes a few years for the effects of an expansion to be felt fully IMHO...

It's not until the Expansion teams start pulling on the rest of the league seriously that they can really have an impact on the run scoring distribution or on the overall average level of competition faced by any one player.

This is a VERY interesting thread. Thanks for taking the time to put together and present this info, Matt.

According to your measures, which are the top 5 strongest and weakest leagues (in both NL and AL) in history?

As to your quote above, though- if this is true (that it takes a few years for the effects of expansion to impact league quality) then why is the spread between the best and the average player (expressed in TRP, Win Shares, whatever) felt most the year of expansion? Why should it take the lousy expansion teams a few years before they start depricating the league strength? That doesn't make sense because expansions teams are typically their worst the 1st or 2nd year, anyway.

I ran a little study where I looked at the comprehensive value metrics pre and post expansion in the early 60's (both NL and AL). On both sides of the ball (esp in the NL), the spread was MUCH greater during the expansion year than the year previous.

Although the term "expansion" obviously needs to be modified, look at the expansion from 1900 (8 teams in the entire Majors) to 1901 (16 teams now in the majors). Nap Lajoie, 1901 is another example. Look at what happened to Cy Young's quality right away, as well. Haven't looked at your results in detail, but if 1901 isn't one of the weakest years for both leagues (ESPECIALLY the NL, as the AL raided their talent, taking 70% of the best with them), then something is clearly awry here.

Why are the records broken during expansion years, after holding up for ~35 years previous? Sportswriters predicted McGwire would break the record in 1998, 2 teams were added (infusing MANY minor leaguer talent caliber players into the NL), and both he and Sosa shattered Maris' mark.

Finally, how do you think your results would correlate with James' league quality assessments delineated in his NHBA? I'm fairly certain he claimed that either the 58' or 59' NL was the most competitve league in history, although his semantic inference regarding the word "competitive" probably differs from yours.

[Ubiquitous]

Well also expansion happened at the same time that the schedule expanded, so there is more wins to go around and more runs to go around. That could be something.

[Ubiquitous]

Mark McGwire hit 58 homers in 156 games and playing two thirds of his games in the Coliseum in 1997. That is why people thought Mark would break the record, he was peaking and he was moving away from the pitchers park to a park that is pretty good for home runs.

[SABR Matt]

To csh...

1901 was indeed a very weak year in baseball history...league difficulty scores of ~.720 existed in the AL for the first few years although it actually dropped lower during the depths of the deadball era (1905-1909)...I suspect this is because they started with a certain level of talent but had not developed the mechanisms to scout NEW talent and as the old guard disappeared it took a few years to develop new quality players.

In the NL the slide is much more rapid...they started 1900 with EDR2s of .729...now it took a couple of years for the slide to take hold, but by 1903 they were below .700 and by 1904 the NL was the weakest it's ever been aside from when it was the NA and when the AA and UA were competing for NL talent between 1881 and 1887 (well OK...there was no AA in 1881 but the league did expand from 8 teams to 12 that year I believe)

I think we need to be careful in assuming that the spread between the best and worst players is specifically what sets the standard for league depth. It's a good hint, but it doesn't fully explain what makes a league hard/easy...a difficult league is one in which a larger proportion of the players fall between a narrower range of skill levels. WS ratings won't really see that kind of information because it's based on team wins...and no matter how "difficult" a league is...the wins are determined by balance between teams...which isn't the same as difficulty. This isn't a semantic issue...this is very real and extremely important...competitive balance does NOT equal league depth...I need to make sure you are clear on that point.

Now...I think what may be causing the delay in EDR2 drops after an expansion is the fact that my skew measurements are centered...meaning on the year of an expansion, I'm seeing the years before it...and the years after...I may try to place hard barriers at points in the history of the game where the league irrevocably changed...and only take skew measurements of data up to those barriers from either direction.

In the AL in 1961 for instance...from 1958 to 1960...the league was 8 teams...those years shouldn't impact the EDR2 of 1961.

I'm going to try this before anyone panicks about delays in feeling the effects of an expansion.

However...there is a logical reason why it takes time for an expansion to impact the run scoring distribution. When a new expansion team emerges...it will have a direct effect on only the games it plays...a small portion of the total games played by the league...no one else will feel those effects. After a few years though...the new team/teams start pulling players out of other franchises...negatively impacting depth league wide isntead of just locally.

Analysts may have correctly predicted McGwire's 70 HR season in 1998...but that doesn't tell me much about the middle of the league...just the extrema...

[SABR Matt]

Mark McGwire hit 58 homers in 156 games and playing two thirds of his games in the Coliseum in 1997. That is why people thought Mark would break the record, he was peaking and he was moving away from the pitchers park to a park that is pretty good for home runs.

True that.

And yes...larger deviations in the RANGE of WS and TPR in 1961/1962 are partially explained by the increase in schedule length...I do think though that you'll see a sharper difference between pre and post expansion seasons if I use a hard barrier where league conditions changed.

[SABR Matt]

Well Ok...at least as far as the AL aroudn the 1961 expansion goes, a hard cap didn't help...it made the 1961 season one of the strongest on record.

I'm not totally convinced that the '61 expansion really did have a negative impact on the quality of the AL...I think it just resulted in the AL pulling more players out of the NL and getting more use out of Negro Leaguers...an expansion will only hurt if you have nowhere else to find your players...

[SABR Matt]

Part of the reason I think I'm on the right track, even if some of the results aren't perfect (there may need to be other adjustments in the future) is that the definition of skew seems to directly apply here.

A larger positive skew is telling you that there's more weight in the run scoring distribution on the extreme end relative to the mean...physically this means that compared to the normal run scoring distribution, seasons where there is larger skew come with more activity in extreme games on iether side of the mean (if the mean doesn't change...and there are a lot of games where a lot of runs are scored..there must be a large number of games where few runs are scored...the reason those don't balance out the skew and leave us simply with a bell curve that has a larger standard deviation is that the mean is a lot closer to zero runs (the fewest you can score) than it is to the largest numbers of runs typically scored...this is why there will always be positive skew. But it stands to reason that more skew means more imbalance in the action from game to game...which means that MOST of the players are weaker (the median of the distribution occurs below the mean by a larger amount) than normal...

If I'm not seeing a larger skew...even in an expansion season...that means that the games more consistant...which to me means that the players were more consistant...than is commonly assumed.

[Ubiquitous]

Players can be consistent and not have high quality. They can be consistently bad across the board as well as consistently good.

I have to go to work but possibly when I come back I'll look into the 1961 AL and see just who was added to that league.

[SABR Matt]

Let me put this question to you guys who are still somehow reading this thread.

Can you think of a way a league could be significantly less deep without a change in the skew of its' run scoring distribution? I'm asking this seriously because if someone can explain to me how that could happen..how there could somehow NOT be more extremes in run scoring relative to the mean and yet have the players be worse...because I can't see how that is possible, but I could be blinded by my own preconceptions and I want to make sure I'm not missing something really obvious.

[SABR Matt]

See, Ubiquitus...I don't really think that's possible..."consistant" badness doesn't happen in nature as far as I can tell...what's the average score of your typical game of slow pitch softball? Chances are it's something like 26-14

That's "bad' baseball...little league...is bad baseball (sorry kids...!)...the GCL is bad baseball (sorry Jose Tabata!)...the 1870s...they were HORRIBLE by major league standards..there's a reason the average RS distribution back then was so spread...and it's not because there were bad players and good players...because there were NO GOOD PLAYERS...except possible Roger Connor and Cap Anson...that's i! They all sucked...HARD...from game to game you never knew what you were going to get. One team managed to score 49 runs in a game in 1871...LOL

I don't think badness can be consistant because it is DEFINED as the inability to play well consistantly.

[csh19792001]

Mark McGwire hit 58 homers in 156 games and playing two thirds of his games in the Coliseum in 1997. That is why people thought Mark would break the record, he was peaking and he was moving away from the pitchers park to a park that is pretty good for home runs.

People knew that a slew of minor leaguers pitchers and fielders were being added to the league, just as had happened in 61' when the league was similarly depreciated. It might have been better than Oakland, but Busch wasn't exactly a good homerun park. If that was a part of the rationale for the belief on the part of the sportwriters, it was likely a much smaller part of it.

Do home run factor data exist for that period? I'd be interested to see how the parks impact scoring.

[csh19792001]

Matt-
Could you list the top 5 (or 10) strongest and weakest years in baseball history for both the AL and NL? I'd like to see how it relates to expansion and other factors external to statistics.

[SABR Matt]

1984 AL
1996 AL
1985 AL
1935 AL
1997 AL
1998 AL
1986 AL
1983 AL
2000 AL
1995 AL

Those are the top 10 years in the AL for strength

1983 NL
2003 NL
2004 NL
1984 NL
2001 NL
2000 NL
1998 NL
1963 NL
2002 NL
1999 NL

Top ten years in the NL

1902 AL
1901 AL
1910 AL
1903 AL
1904 AL
1908 AL
1909 AL
1905 AL
1906 AL
1907 AL

Bottom ten years in the AL

1876 NL
1877 NL
1879 NL
1886 NL
1884 NL
1883 NL
1882 NL
1881 NL
1885 NL
1880 NL

Bottom ten in the NL

[Ubiquitous]

See, Ubiquitus...I don't really think that's possible..."consistant" badness doesn't happen in nature as far as I can tell...what's the average score of your typical game of slow pitch softball? Chances are it's something like 26-14

That's "bad' baseball...little league...is bad baseball (sorry kids...!)...the GCL is bad baseball (sorry Jose Tabata!)...the 1870s...they were HORRIBLE by major league standards..there's a reason the average RS distribution back then was so spread...and it's not because there were bad players and good players...because there were NO GOOD PLAYERS...except possible Roger Connor and Cap Anson...that's i! They all sucked...HARD...from game to game you never knew what you were going to get. One team managed to score 49 runs in a game in 1871...LOL

I don't think badness can be consistant because it is DEFINED as the inability to play well consistantly.


Take the 30's AL as an example. It was consistent (there is an "E") throughout much of the decade and it ends up coming out better then the 50's, 60's, and 70's. Now is that because the 1930's AL had better league quality and depth then those decades or was it simply becuase they were more stable while in the other periods it was going through a transistion? Because of the transistion the mechanics get screwed up and it looks like the league is lower quality when in reality the league is better but because it is going through a transition it has some peaks and valleys. Whereas the 1930's AL was stable. I guess the best example I can give is comparing a 10 year old and a 15 year old. A 15 year old is going through lots of changes, growth spurts, and so forth. So most of the time he appears a mess, while the 10 year looks relatively the same with a normal growth rate. Comparing that 10 year old to other 10 years olds will probably show that the age group is relatively close together in most things. While the 15 year old group is probably all over the board. But that doesn't mean that the 10 year olds are the better group. The 15 years old with all the mess and all the different rates are still better then the 10 year old physically and mentally. Understand at all.

[Ubiquitous]

People knew that a slew of minor leaguers pitchers and fielders were being added to the league, just as had happened in 61' when the league was similarly depreciated. It might have been better than Oakland, but Busch wasn't exactly a good homerun park. If that was a part of the rationale for the belief on the part of the sportwriters, it was likely a much smaller part of it.

Do home run factor data exist for that period? I'd be interested to see how the parks impact scoring.


Yes the data exists for those periods. I'm at work but overall Busch stadium is a better then average home run park, and a much better park for homers then the Coliseum.

[SABR Matt]

My research into park effects suggests that the negative effects of the collesium are GROSSLY overstated because it is being compared only to the rest of the AL...in fact of the 14 parks in the AL, I now believe 10 are hitter's parks...3 are neutral, and 1 is a pitcher's park...standard park statistics will COMPLETELY miss this.

[SABR Matt]

Stability of who's in the league shouldn't have an effect on the skew.

If the level of play stays fixed, the skew won't CHANGE much (that's not actually what happens by the way...the AL had a glut of brilliant position players on the better pitchers were there too back in the mid thirties...and oh BTW..the skew changed a LOT in that decade...dropping dramatically and then rising dramatically as the decade began to turn)....but a bad league should have a broad range of run scoring outcomes...consistantly so.

[Sultan_1895-1948]

My research into park effects suggests that the negative effects of the collesium are GROSSLY overstated because it is being compared only to the rest of the AL...in fact of the 14 parks in the AL, I now believe 10 are hitter's parks...3 are neutral, and 1 is a pitcher's park...standard park statistics will COMPLETELY miss this.

Not sure if this is the case or not, but:

Isn't it possible that every park is a hitters park in the grand scheme of things. I mean, with park factors, there HAS to be a some "pitchers" parks, because they are all being compared with eachother. It only makes sense that some will yield more offense than others, but they all are hitter friendly.

[SABR Matt]

Huh?

Well I understand the point you're trying to make...but the point of park effect analysis is to determine whether relative to the environments of the other players in baseball...a player who spends his time at any given park is being given an advantage or disadvantage.

The American League has in recent times skewed toward having the smaller parks in baseball and the NL has the larger parks...in fact we estimate that the NL parks are allowing about 0.3 runs per game per side FEWER than the AL parks all by themselves...with other contextual factors removed over the last five years.

This is causing the pitcher friendliness of parks like Comerica, Yankee Stadium, the Oakland Collesium and recently...the revised Kauffman stadium to be grossly exaggerated while parks like Minute Maid, Miller Park, and Bank One...each a slight hitter's park in reality...are being made to look like homer-havens...

For further elaboration on the method that produced these findings see the thread in sabermetrics here posted by me last month that featured the Fiato-Souders Arithmetic Adjustments Matrix

[SABR Matt]

Minor correction...that information came to light in a thread entitled "Babe Ruth's Top Five Seasons"...it's on I think the second page of that discussion where we brought out the FSAA Matrix in an effort to settle the question of how homer-friendly the Polo Grounds/Yankee Stadium were.

[BillyF29]

So, with adjustments made and all that jazz, what do you think are Babe's top 5 seasons.

I have:

1. 1923
2. 1921
3. 1920
4. 1924
5. 1926

with 1927 and 1928 very close behind.

And the top 10 seasons since 1900:

1. Barry Bonds 2001
2. Barry Bonds 2004
3. Babe Ruth 1923
4. Mickey Mantle 1957
5. Babe Ruth 1921
6. Mickey Mantle 1956
7. Barry Bonds 2002
8. Ted Williams 1946
9. Honus Wagner 1908
10. Mickey Mantle 1961

[SABR Matt]

Well it appears the AL was a tougher league in '23 than it was in '20/'21. And if you include our new linear/arithmetic park adjustments...it's a little closer still...I think it can go either way between '21, '20 and '23...talent wise it's probably:

'23
'20
'21
'27
'24

[Sultan_1895-1948]

Fiato-Souders Arithmetic Adjustments Matrix

Huh? :lookitup :eek:

[Sultan_1895-1948]

Well it appears the AL was a tougher league in '23 than it was in '20/'21. And if you include our new linear/arithmetic park adjustments...it's a little closer still...I think it can go either way between '21, '20 and '23...talent wise it's probably:

'23
'20
'21
'27
'24


'21 has always been my favorite, which is a completely different issue than which season was a tougher league. Leaving Saber out of it, I'd go '21 as his best.

[SHOELESSJOE3]

'21 has always been my favorite, which is a completely different issue than which season was a tougher league. Leaving Saber out of it, I'd go '21 as his best.

I like 1921. It was the year he put it all together.
Best year for RBI's 171
Best year for runs scored 177
Best year for triples 16
Total bases 457
EBH's 119
Second best year for slugging
Second best for doubles
Second best for Batting average
Second best ISO
Second highest number of hits 204
Second best for home runs
On the home runs only a fluke, fan interference took away his first 60 home run season.July 6,1921 Ruth hit a ball into the right field stands and a fan attempting to catch the ball knocked back on to the field. The Yanks protested, lost the argument, awarded a double.

BTW that year 1921 that he hit his second best total of doubles it was only one less than his best, his slugging was only one point less than his best, total hits only one less than his best, home runs only one less than his best.

One can debate the league strength in different years but this was his best all around season.

[misterdirt]

See, Ubiquitus...I don't really think that's possible..."consistant" badness doesn't happen in nature as far as I can tell...what's the average score of your typical game of slow pitch softball? Chances are it's something like 26-14

Sounds like you think the very best ball players are not in the major leagues, but are playing FAST pitch softball where the average score is pretty close to 1 - 0.


Can you think of a way a league could be significantly less deep without a change in the skew of its' run scoring distribution?

See above.

Also, take a situation like the integration of major league baseball. If a few teams integrate while others, the result should be wider variations in scores, which under your method, if I understand it correctly, would indicate a weaker league. But the variations in score would have been induced by ADDING to the league from a pool of high quality players previously unavailable, so the total overall quality of the league should have increased. It seems like what you are measuring is league balance, which would be affected by the relative economic balance of the teams as much as the total overall quality of the players.

[SABR Matt]

Fast pitch softball...from the games I've seen...doesn't average close to 1-0...but those scores are somewhat more reasonable...(I've seen only a handful of games though so I could be wrong...they were always 7 inning contests and wound up with final scores like 6-3 or 5-1)

Here's a question for you...how would *you* go about attempting to measure league depth? (I'm actually very curious here) If league skew measures balance...why doesn't it show a drop in balance during expansion years? And if skew in RS per game doesn't measure skill...what would?

I'm hoping for some original inspiration, because apart from using data that's a little oversimplified like the rate of unearned runs scored or something like that...I can't think of another approach.

[misterdirt]

I was refering to top quality men's fast pitch leagues like the traveling teams based in and around NYC.

I appreciate your wanting my opinion, but I have to confess that it is a problem that I have not thought too much about since, as you know, I am not at all interested in ranking the past accomplishments of players. Your method seems as good as any but since I know that it is important to you to get it as accurate as possible I thought I would point out what I thought were some potential pitfalls for you to consider. (Not the fast pitch softball stuff but the other comments.) If another method occurs to me I will let you know. Best of luck.

[Blackout]

The American League has in recent times skewed toward having the smaller parks in baseball and the NL has the larger parks...in fact we estimate that the NL parks are allowing about 0.3 runs per game per side FEWER than the AL parks all by themselves...with other contextual factors removed over the last five years.


you dont think the DH has anything to do with those 0.3 runs?

[Ubiquitous]

.

Also, take a situation like the integration of major league baseball. If a few teams integrate while others, the result should be wider variations in scores, which under your method, if I understand it correctly, would indicate a weaker league. But the variations in score would have been induced by ADDING to the league from a pool of high quality players previously unavailable, so the total overall quality of the league should have increased. It seems like what you are measuring is league balance, which would be affected by the relative economic balance of the teams as much as the total overall quality of the players.
Thats basically the point I was trying to make a few posts ago.

[SABR Matt]

you dont think the DH has anything to do with those 0.3 runs?

No...as a matter of fact I don't. I understand why you would say that...Randy and I had the same concerns...however, if you actually take a look at all of the parks in each league...it becomes very evident that the parks are in fact very much different from the AL to the NL. The DH...from what we can tell...since 1993...has explained about 1/3 of the margin between the AL and the NL RS counts...the parks have explained the vast majority of the rest (the league gets credit for being slightly more favorable than the NL most of those years...but not by much...which makes sense...it's not intrinsic that the AL should be more hitter friendly unless they're using different balls in each league or the umpires have league biases...which could explain the minor differences)...

Here's the thing...where in the NL...the pitcher and his subsequent replacements bat somewhat worse than the AL DHs...it's not as large a spread as you might think because pitchers only take about 1/2 of those at bats...and because as it turns out...a lot of AL teams have CRAPPY DHs. And the real trick...in the AL, because you have the DH, teams feel comfortable starting Pokey Reese, John Olerud, Jeremy Reed, Yuniesky Betancourt, David Segui...etc...a whole host of sucky hitters that don't start in the NL...they get to play the field...that's why fielding is better in the AL. So although the DH does improve offense some...it also improves defense and the offensive gains are not what you might expect.

[SABR Matt]

Thats basically the point I was trying to make a few posts ago.

I believe I understand what you're saying...if interleague only took hold in a handful of the teams...then those teams would be disproportionately good and would theoretically dominate the run scoring distribution.

I wonder if there's a way to adjust for the intrinsic imbalances in teams so that the only difference in skew I'm seeing is related to the inequity in player depth over the entire league.

[SABR Matt]

OK...one moment here...I just remembered I tested to see if there was correlation between the skew or standard deviation of team winning percentages and the skew of the run scoring distribution...there was none. Of course I test individual seasons, so W% may be too flaky...I may need to put both the skew of the RS distribution and the standard deviation of team W%s into seven year contexts and see if that clears out some of the noise...

I just can't believe that could result in going from a slight negative correlation (-0.0221) to the kind of significant positive correlation I would need to prove that imbalance in teams causes high skew the way you've porposed.

[SABR Matt]

WHOA...flash of insight...

I think it is anyway...

I think the reason leagues appear to get stronger when there's an expansion is that as you introduce a larger volume in your schedule (more games) you introduce a bias toward smaller skew!

According to probability theory, larger Ns (sample sizes) lead you toward normal...it's the central limit theory...it applies...ALWAYS...I canNOT believe I didn't think of this before...I'm stunned...

And I have work to do...I'm going to be attempting to filter out that bias.

[SHOELESSJOE3]

No...as a matter of fact I don't. I understand why you would say that...Randy and I had the same concerns...however, if you actually take a look at all of the parks in each league...it becomes very evident that the parks are in fact very much different from the AL to the NL. The DH...from what we can tell...since 1993...has explained about 1/3 of the margin between the AL and the NL RS counts...the parks have explained the vast majority of the rest (the league gets credit for being slightly more favorable than the NL most of those years...but not by much...which makes sense...it's not intrinsic that the AL should be more hitter friendly unless they're using different balls in each league or the umpires have league biases...which could explain the minor differences)...

Here's the thing...where in the NL...the pitcher and his subsequent replacements bat somewhat worse than the AL DHs...it's not as large a spread as you might think because pitchers only take about 1/2 of those at bats...and because as it turns out...a lot of AL teams have CRAPPY DHs. And the real trick...in the AL, because you have the DH, teams feel comfortable starting Pokey Reese, John Olerud, Jeremy Reed, Yuniesky Betancourt, David Segui...etc...a whole host of sucky hitters that don't start in the NL...they get to play the field...that's why fielding is better in the AL. So although the DH does improve offense some...it also improves defense and the offensive gains are not what you might expect.

I think the DH does make a difference, more than you believe. The problem is, how do we measure, detemine hiow much of that difference is the parks and how much is effected by the DH.

Some figures for the period 1980-2004

National League-----------Batting Ave.------OBA
All batters-----------------.259-------------.327
pitchers excluded----------.266-------------..336

Now, I must admit, I jumped right into this one. Have not given it much thought, there may be some flaws, example there is a much greater number of at bats over all those years when pitchers at bats are included. I will look and think it over but on the face it's apparent that it would be more favorable to have DH in the batting order than a pitcher.

Just dug this bit up.

Over that period 1980-2004.
NL pitchers batted----------.145---------.180 OBA
AL DH batted---------------.265---------.345 OBA

[SABR Matt]

I think you'll find shoelessjoe...that when you sum up all batters who were either a pitcher or one of the myriad pinch hitters who batted for the pitcher...the numbers will be somewhat less horrid...I think you will also find that although AL DHs are obviously better offensively than the pitcher/pinch hitter circle and of course...get many more at bats than the guys hitting 9th in the NL...that two things about the way AL and NL baseball are played differently minimize the impact of the DH

First...there are more defensive specialists getting more PT in the AL than the NL.

Second...NL line-up trips are planned to work around the pitcher...national league teams don't count on every line-up spot to do their run scoring...they work situationally to avoid as much as possible needing anything out of the pitcher other than a sac bunt every now and then.

[Ubiquitous]

First...there are more defensive specialists getting more PT in the AL than the NL

Dusty Baker would disagree. Any actual proof of this? What is the STD of players offensive skills? How many at bats do "defensive specialists" get in the AL? How many in the NL? What is the floor in the NL? What is the floor in the AL?


Second...NL line-up trips are planned to work around the pitcher...national league teams don't count on every line-up spot to do their run scoring...they work situationally to avoid as much as possible needing anything out of the pitcher other than a sac bunt every now and then

Lineup construction outside of the first inning is pretty pointless and what works and doesn't work optimally has been figured out a long time ago. Its not like AL teams are putting their worst hitter in the three spot as opposed to in the back of the lineup more so then that is happening in the NL. The AL doesn't expect much out of their 7-8-9 hitters either.

[SABR Matt]

a) I'm not claiming the DH has had no effect...just that the actual impact has been somewhat exaggerated.

b) No I haven't done point by point research as of yet (I of course intend to)...to determine how many at bats "defensive specialists" get in either league (not sure how you would define that exactly)...I think perhaps the best way to go about it would be to look at the distributions of a good offensive metric...i.e. NOT batting average or OPS or anything lame like that...a Runs Created metric of significant value)...to see if those distributions look similar but with one of them shifted (indicating a league bias outside the control of the players like park effects) or if there's a real difference in the shape.

[antihipster]

Recently, I have been running an era adjustment test run on the pitching k:bb ratio. In general, I have found that as the years go along, there is less of a skew, even though some outragously great pitchers can break this rule [ie, Bob Gibson, Randy Johnson]. So far, the largest skew is for Tommy Bond, [as a lot of pitchers from this era also get this recognition.]

The method I used is starting out w/ a player's career k:bb ratio. Then I get the league K:bb ratio and subtract the player's k:bb from the league total. At this point, I divide both the league and players k:bb. Then I add the league bb:K [which includes the player being calculated] and do the same thing with only the player's bb:k results. Finally, the player's k+bb is divided by the league's sum of K+bb, and I subtract this from the player's K:bb percentage, which is then subtracted from the league average.


I thought the history buffs might find this one interesting enough that I decided to post it here...the "sabermetrics" involved here are very light mathematically, so it fits in.

This is just experimental, because to properly scale my difficulty rating, I had to arbitrarily choose the marginal value you'll see in a moment...I'm working on ways to more rigorously define it.

OK...bear with me for a moment while I explain where I got this idea.

I've been looking for the LONGEST time for a way to objectively rate how "deep" or "difficult" a league was...

I never liked James' subjective timeline adjustment...it seemed WAY too simple. But how do you go about seeing how skilled the players within a league are as a group?

The idea came to me through a discussion I had with Randy Fiato (TKD) about what defines "bad baseball". It is intuitively obvious that when two bad teams face each other, the games will be sloppy more frequently..mistakes will be made in all aspects of the game. Pitching mistakes...hitting mistakes...fielding mistakes...baserunning blunders.

What will this look like statistically though? A classic idea proposed by sabermetricians in the 70s was to rate players based on standard deviations from the mean...it has been observed many times that the standard deviation of batting average has been fluctuating through time but trending down...(there's a famous paper on the disappearance of the .400 hitter that discusses this...the author's name escapes me for some reason).

Batting average is not however explanatory enough...what we want to know is...does the standard deviation of run scoring per side per game change with time the way it does for batting average? Are we cycling closer and closer to the mean as time advances?

A quick survey using retrosheet.org's game logs reveals that in fact standard deviation is changing with time...but perhaps not the way you might think. It became immediately apparent that the standard deviation of run scoring on a per game basis was directly dependent on the league average run scoring rate. In fact, an r^2 of 0.9301 exists between those two variables...low scoring leagues have small standard deviations...high scoring leagues have larger standard deviations.

Does this mean that high scoring leagues are "weaker"...less deep with talent? Of course not. It's hard to argue that the deadball era was a better level of play than today's game even with expansion considernig the player pool has expanded to include approximately 50 times more potential baseball players than it did back then, minor league scouting and development didn't exist in the deadball era, and the equipment and field conditions were often horrendous, making for sloppy games far more frequently than in today's major leagues.

This dependence on run scoring environment is not however the only problem with using standard deviation to rate the difficulty of a league or the players within the league. There is a fundamental logical flaw. The use of standard normal z scores presumes that the league and/or player distribution was normal...neither is the case.

The player distribution is pyramidal...the top 1% of the humans who play baseball make the major leagues (liberally...it might be closer to .001%)...if we could rate every baseballer from tee-ball to Japan to MLB to High School...the distribution of skill might be normal. Meanwhile, the distribution of runs scored per side per game in a league is the summation of a series of one-game match-ups...each match-up behaving according to the laws of probability as governed by the intrinsic strengths of both combatants...the result of that process is a non-normal significantly skewed distribution...high extreme values will have an exaggeratedly large Z-score...shutouts are a sign of bad play too but their is a lower bound to how "bad" you can be in the non-scoring direction.

Given this lower bound...and the resulting tendency for variations in ability to manifest themselves in the rightward biasing direction (large numbers of high scoring games relative to the mean run scoring environment)...we fall back on MEASURING the skew of the league's RS distribution to get an idea about how erratic/weak that league was.

The positives...Skew is not dependent on the run scoring environemtn...it is never affected by the mean of a probability distribution. Skew uni-directional...meaning the lower bound shouldn't interfere with an accurate measurement of positive skew (skew is defined to be positive when the longer tail of a distribution points to the right on a number line). Skewness also does not presume a distribution is normal. It describes how non-normal a distribution is.

Logically...skew tells you how frequently extremes occur...more extremes mean more variation in intrinsic team strengths...and therefore...a weaker league.

If the run scoring distribution were normal (had no skew) this would mean that there was ZERO variation in player ability across the league...this would be the "ideal" league...but we know this to be humanly impossible to achieve...nonetheless...it serves to demonstrate that more skew is a larger deviation from the ideal league.

Skewness of a distribution is easily measured:

SUM(x - u)^3
--------------------
(n - 1) * s^3

Where x is the observed game/side runs scored, u is the league average runs scored per side per game, n is the number of game/sides within the league and s is the standard deviation of the distribution.

Placing the s term in the expression automatically scales the skew value so that higher scoring leagues, which will naturally have a wider range of run scoring outcomes do not appear to have higher skew.

When I plotted skew of the run scoring distribution against time, wat I found was a somewhat messy but nonetheless encouraging trend toward gradually decreasing skew with time. There was a lot of noise in the plot...probably because skew is heavily impacted by large outliers, so extreme games might have had a disproportionately large pull on skew...it therefore was necessary to smooth skew values.

I chose to use a normally weighted 7-year running mean of skew values for each league (normally weighted implies a larger emphasis on the center year...think of the shape of the bell curve) to smooth out the fluctuations...

It makes sense to smooth the data because although players change from season to season...the overall strength of the league cannot possibly fluctuate by overly large amounts...there are hundreds of players in any given league...turnover from year to year is no larger than 5-10% so we should expect league strengths to change gradually except in extreme circumstances like during WWII.

I'm considering alternatives to this normally weight running mean idea...I may for instance measure the skewness of a longer period of years than one...perhaps skew is more persistant if you incluide more than one year of data...either way...the smoothed values were eye popping and aligned very well with my expectations for where baseball was weak and where it was strong.

But this doesn't end the problem.

Assuming Smoothed skew is an appropriate measure of league strength, we need to put it in a form that allows strong leagues to score higher than weak leagues...and it would in fact be ideal if we got the scores to range from 0 to 1 so that they could be used multiplicatively...(for instance...if we rate 1872 as a 0.5 league...we would cut player wins in half in 1872 to get an idea of how many wins they'd be worth in a strong league)

We can make use of the exponential function here...it makes sense to use the exponential given that major league baseball represents the top of the baseball pyramid and the drop in skew value from typical leagues to great ones is likely to be large.

It also gives us the right range if used properly. Skewness can theoretically range from 0 to infinity in this case (it can't range negatively because of the lower bound at zero)...if we take a skewness of zero...e^0 = 1...if we take a skewness value approaching infinity e^large = large...ah but if we make that e^-skew...-0 is still zero, but -large implies 1/(e^large) which asymptotically approaches zero.

One more step though...no baseball league...no matter how great...will ever have a skew of zero. Here's the nasty part where I have to arbitrarily pick a marginal skew value. This was just me visually examining the graph of smoothed skew with time and seeing what the skew appeared to be approaching (the overall curved trend appears to be leveling off slowly but surely.

I chose a value orf 0.8 as the minimum skew...though I experimented with other values.

This was applied by simply subtracting 0.8 from each skew value obtained by the smoothing process before converting them with the exponential decay function.

The end result is quite interest to me...

Here are the top 20 most difficult leagues by this method:
Year Lg Strength
1984 AL 0.968
1985 AL 0.967
1997 AL 0.947
1995 AL 0.946
1996 AL 0.943
1998 AL 0.942
1986 AL 0.941
1983 AL 0.941
1983 NL 0.932
1933 AL 0.928
1934 AL 0.928
1999 AL 0.925
1994 AL 0.925
1982 NL 0.923
1937 AL 0.919
1935 AL 0.913
1938 AL 0.912
1936 AL 0.909
1987 AL 0.907
1962 AL 0.906

And the 20 weakest leagues
1910 NL 0.691
1909 AL 0.690
1944 NL 0.688
1902 NL 0.687
1901 NL 0.683
1885 NL 0.682
1905 AL 0.679
1911 NL 0.675
1881 NL 0.666
1875 NA 0.665
1906 AL 0.663
1908 AL 0.654
1907 AL 0.651
1874 NA 0.637
1873 NA 0.614
1884 NL 0.612
1882 NL 0.589
1872 NA 0.578
1883 NL 0.560
1871 NA 0.528

The early deadball era looks to me to have been very weak competitively...though obviously not as bad as the old National Association...which plays like a modern AA or A league.

Thoughts from the peanut gallery?

[iPod]

1984 AL was the strongest? So that would make the '84 Tigers the best team ever in a landslide?

[SABR Matt]

Not quite that simple, but yes, that lends added weight to the success of the '84 Tigers

[SHOELESSJOE3]

I think you'll find shoelessjoe...that when you sum up all batters who were either a pitcher or one of the myriad pinch hitters who batted for the pitcher...the numbers will be somewhat less horrid.

Probably true. I gave only batting averages for pitchers and those figures, stats of course do not include at bats, batting averages or OBA of pinch hitters who did bat for pitchers. Those numbers and comparisons that I posted can be viewed in post 91#.

So that may narrow the gap between what the pitchers in the NL compared to the American League DH do if we were to include the results of the pinch hitters who pinch hit for the pitchers.

How much it narrows the gap hard to say but keep in mind I would think on average the pitcher would have at least two at bats in most games.

[SABR Matt]

Most positions on the diamond get 4 at bats per game...there were about 2600 games in the NL last year and the pitchers got about 5000 ABs...that's about half the at bats taken by the 9th slot in the order...maybe a little more.

Which tells me that assuming the other half was roughly league average...you can chop the deficit between NL Pitchers and the other NL position players in half...there's still a significant difference between the AL DH and the NL P/PH group, but it's not quite the dramatic gap people envision.

[Ubiquitous]

In 2004 the pitchers of the NL had about 5850 plate appearances. The DH's of the AL had about 6,900 Plate Appearance. DH's are worth roughly 895 runs per 5,500 plate appearances, and the pitchers are worth 188 runs.

Pitchers line: .147/.181/.189
DH line: .270/.349/.466

[SABR Matt]

So let's see if I follow here...

we're talking about approximately 600 runs distributed over 2600 ballgames in the NL that are lost by batting the pitcher and his cadre of pinch hitters compared to the AL DH slot.

600 runs in 2600 games...that's about .23 run per game lost in the NL...it doesn't fully explain the difference between the two leagues...not in the 90s at least...which makes sense...the AL parks are more hitter friendly than the NL parks...they're adding significantly to the offense in the AL.

Moreover, I don't think you can make the case that the AL's other hitters besides DHs are as good as the NL's other hitters besides pitchers/PHs. I think there are more defensively gifted first basemen, and left/right fielders who don't hit as much in the AL and more power hitters at the corers in the NL...hey ubiquitus...can you run the league numbers without pitchers/PHs and without DHs?

[Ubiquitous]

In 2004 the NL without pitcher is .270/.341/.437

For the AL since I was never very good at Access:

Posi AVG SLG OBA OPS AB
Outf .278 .441 .344 .785 27546
Thir .265 .437 .338 .775 9778
Shor .272 .417 .326 .743 8760
Seco .264 .407 .328 .735 9225
firs .266 .452 .349 .800 8826
Catc .263 .416 .326 .742 8267



I have the PBP data for 2004 but have never really figured out Access. Have A.S.S. for ealier years, but have to go to work. Perhaps I'll crunch those numbers afterwords.

[Ubiquitous]

Here is some tidbits for you.

In 2000 the AL pinch hitter batted .249/.343/.354
In the NL it was .226/.319/.348, the pitcher .150/.186/.195
The AL DH batted .276/.361/.463.

The NL pitcher and PH hitter came to bat almost as much as the american League DH and PH.

The AL without the DH or the pitcher hitting had this line:
.276/.348/.442
NL: .273/.350/.446

Everything included:
AL: .276/.349/.443
NL: .266/.342/.432

It looks to me like the average AL-NL positional player is about equal. Why? Well once we adjust for things like Coors field and such I think they become rather close. For instance I looked at years before Colorado came in and they are very close while the data I have for 1999-2002 has the NL with a slight OBP and SLG advantage.
It looks to me at least if the park factors work the way I think they will that the AL does not employ weaker hitting players because of the DH. It looks to me like the AL employs the same kind of positional players plus another hitter.

[SABR Matt]

See...I don't buy your assertion that we should factor in Coors Field and somehow ignore the hitters parks in the AL...in fact prior to the walls being pushed back 15-20 feet in all directions, Kauffman Stadium was ALMOST AS BAD AS COORS...but...whatever you say. :\

I think you're making a classic mistake...you're assuming that all things else are equal and comparing those batting lines as if they could be compared directly. They CAN'T...the average AL Park is IMHO significantly more hitter friendly currently than the average NL park...

[SABR Matt]

More work today on the Era Difficulty Rating...

I've done several things here.

First and foremost, I decided that fluke games were partially responsible for the erratic nature of the individual season Skew scores...if you get one 27 run game/side...(it's happened several times even in modern seasons)...it can have a HUGE impact on the skew for that season.

What I've chosen to do to combat this problem is reanalyze the RS distribution with the top 1%, 2.5%, and 5% of the game/side scores removed...this means I had to recalculate mean, standard deviation and skew to fit the new distribution.

I also realized I did not properly weight seasons when finding 7 year normally weighted skew scores...I should have weighted those seasons by games which was not done. So the new weighting scheme includes the correct normal weights as well as a factor for games played in each season...this will help de-emphasize findings based on very short seasons like 1994, 1981, and of course much of the 19th century.

I smoothed the non-trimmed distributions, the 1%, 2.5%, and 5% trimmed (and here...when I say I trimmed a certain percentage...that's all coming off the right hand side...to elimiate extreme blow-out games) distributions and found EDRs assuming that the current best single season skew value (best meaning lowest) was the skew floor (thus the best season gets a 1 EDR because EDR = e^(-1*(Skew - Skew Floor)).

The result is a set of 4 EDRs for each league and season (non-trimmed, 1%, 2.5%, and 5% trimmed)...the more trimming I did, the more stable skew became..and the less it tended to decrease as you moved forward in time.

I took these EDRs and averaged them together for each league/year entry to come up with a net league/year EDR

I also averaged who years together (blended all of the leagues in that season) to get a single number for the entire year...

When I did that I got a REALLY fascinating repeating pattern in the graph of EDR with time.
Year TEDR
1871 0.713
1872 0.711
1873 0.718
1874 0.725
1875 0.725
1876 0.729
1877 0.755
1878 0.771
1879 0.777
1880 0.780
1881 0.776
1882 0.792
1883 0.779
1884 0.806
1885 0.808
1886 0.803
1887 0.790
1888 0.792
1889 0.812
1890 0.845
1891 0.835
1892 0.866
1893 0.893
1894 0.903
1895 0.898
1896 0.881
1897 0.863
1898 0.854
1899 0.850
1900 0.835
1901 0.818
1902 0.798
1903 0.779
1904 0.765
1905 0.769
1906 0.773
1907 0.773
1908 0.781
1909 0.799
1910 0.816
1911 0.829
1912 0.847
1913 0.867
1914 0.866
1915 0.847
1916 0.841
1917 0.836
1918 0.841
1919 0.853
1920 0.869
1921 0.880
1922 0.878
1923 0.870
1924 0.876
1925 0.889
1926 0.900
1927 0.904
1928 0.895
1929 0.882
1930 0.872
1931 0.873
1932 0.876
1933 0.875
1934 0.874
1935 0.874
1936 0.881
1937 0.896
1938 0.907
1939 0.904
1940 0.898
1941 0.889
1942 0.866
1943 0.839
1944 0.824
1945 0.827
1946 0.845
1947 0.870
1948 0.883
1949 0.885
1950 0.882
1951 0.881
1952 0.873
1953 0.867
1954 0.866
1955 0.870
1956 0.877
1957 0.889
1958 0.904
1959 0.922
1960 0.934
1961 0.937
1962 0.930
1963 0.918
1964 0.908
1965 0.899
1966 0.883
1967 0.861
1968 0.855
1969 0.871
1970 0.884
1971 0.882
1972 0.875
1973 0.880
1974 0.896
1975 0.912
1976 0.916
1977 0.913
1978 0.905
1979 0.903
1980 0.915
1981 0.938
1982 0.964
1983 0.978
1984 0.979
1985 0.968
1986 0.952
1987 0.925
1988 0.894
1989 0.874
1990 0.873
1991 0.886
1992 0.897
1993 0.910
1994 0.930
1995 0.942
1996 0.942
1997 0.943
1998 0.941
1999 0.934
2000 0.927
2001 0.922
2002 0.924
2003 0.932
2004 0.945

kinda hard to visualize it in numeric form, but it pulses at regular intervals and in EXACTLY the same way...or at least...very nearly exactly. Some event triggers a talent collapse...the EDRs plummet...they rebound rapidly to a point...appear to stabilize then take off upward again to a new peak...and then the cycle starts again...only with each cycle, the peak gets higher and the trough less severely low...there is one notible exception to my perfect (and COOL!) pattern...and that is 1925-1930...evidence suggests that right about in here there should be a rapid drop and it should pulse back up in the late 30s...this doesn't happen...it stays level through the 20s and has the expected peak in the late 30s.

Could this lake of a downphase in the 20s be caused by the false floor...the lack of minority ballplayers and thereby the unilateral weakennig of the entire league structure? The NL by itself *DOES* show the dip in the mid 20s and peak in the late thirties right on cue...the AL does not.

BTW if the trend continued...we're beginning to climb out of the short stability period between rapid increases in league difficulty...in the next six years, one might expect the league to get extremely difficult again.

[Ubiquitous]

See...I don't buy your assertion that we should factor in Coors Field and somehow ignore the hitters parks in the AL...in fact prior to the walls being pushed back 15-20 feet in all directions, Kauffman Stadium was ALMOST AS BAD AS COORS...but...whatever you say. :\

I think you're making a classic mistake...you're assuming that all things else are equal and comparing those batting lines as if they could be compared directly. They CAN'T...the average AL Park is IMHO significantly more hitter friendly currently than the average NL park...


Kauffman stadium had a few bloop years that were pretty much out of whack with the rest of its history. Coors field and playing in Colorado is a consistent effect that effects all teams and players for the most part. It isn't simply the home team that scores more runs but all teams. It inflates the run total much more then any other park out there. I'm not saying we should ignore anything, what I am saying is that Coors field dwarfs anything the AL can put out there. For every hitters park you name in the AL I can name one for the NL and still have Coors field as the trump card.


Part of the problem is that Coors field is such an impact that it changes what an average park looks like in the NL, makindg some parks look average or slightly below average when they really are not.

The only true Hitters stadiums I can think of in the AL is Fenway and Arlington.

In the NL you have Wrigley, Coors Field, Enron, Citizens Bank Park, and Bank One Ballpark.

[SABR Matt]

Well that's just flat out not true...

Let's run down the list of hitters parks in the AL

Fenway (consistantly strongly a hitters park)
Skydome (very consistantly a hitter's park)
Comiskey (I and II both consistantly favor hitters)
HHH Metrodome (also consistantly hitter friendly)
Kauffman Stadium (before the walls moved out)
TBIA (duh)

In the NL

Coors Field
Bank One Ballpark
Possibly Minute Maid
maaaaybe Citizen's Bank but that's far from certain

Now lets look at known pitchers parks

AL

Safeco Field
?????

NL

Petco and its predecessro in SD
Dodger Stadium
Candlestick/SBC
PNC Park
Great American
Shea Stadium
Pro Player
Turner Field/Fulton County
and of course this year you have RFK and in Montreal Stade Olympique


I rest my freakin' case.

[SABR Matt]

BTW I find it astounding that you of all people could be suckered in by the belief that Wrigley is a hitter's park...it's so blatantly neutral (linked predominantly to the prevailing wind in any given year) as to be hilarious.

I also find it funny that you focused on the hitters parks in the NL and didn't realize the pitcher parks...KNOWN PITCHER'S PARKS in the NL absolutely DWARF the "pitcher's parks" in the AL...Safeco and I guess you could make a case for Yankee Stadium most seasons...that's it...no...not Comerica...that's played nearly perfectly neutral since May of its' first inception year...no...not Macafee/Colleseum...that's also played roughly neutral despite the huge foul territory...

[misterdirt]

Matt - I think you should find this link relavant. I don't agree with his analysis of home run hitting for reasons that I give in a comment on his site but his use of a mathematical analysis to evaluate a non guassian distribution that is highly skewed in a baseball context should interest you. http://www.arthurdevany.com. You want to go to his research paper on why home run hitting hasn't changed.

[Ubiquitous]

blatantly neutral? Wrigley is not blatantly neutral. Wrigley Field for years and years and years was a hitters park. then in the 90's other parks come on line and then everybody is saying how the wind makes it neutral. Huh? The wind didn't exist in 1978? It's still a hitters park just not a Coors Field.


You can rest your freakin case but how is Great American Ballpark a known pitchers park? How is Turner Field a known pitchers park? The GAp has been open 3 years It had one neutral, one pitchers, and one hitters. Yet somehow you think you can rest your freakin case on that one? Turner Field has been open for 9 years and had two seasons in which its park factor was favoring the pitcher yet somehow you can rest your freakin case on it. These two parks you think are locks because they favor your view yet a park like citizen bank is too early to tell. PNC park has been around for 5 years and in that time it has hovered around league neutral yet you somewho see fit to declare it a pitchers park. Yet Wrigley looks similar to PNC the last 5 years but has a higher high is not a hitters park.

Again though the NL bottom is going to look a lot worse then the AL bottom because of Coors Field. I would love to see what the park factors for the NL would look like in the last 10 years without Coors Field.

[Ubiquitous]

Who plays in the NL west?

Colorado
LA
SanFran
San Diego
Arizona.

The majors now use a unbalanced schedule, so now these teams play a Coors field heavy schedule. Guess What? surpirse 3 of the 4 other teams play in a park that looks like very favorable pitchers parks. Gee is it any wonder when they play over 10% of their away games at Coors Field? Only one other stadium manages to hold its head up above the hitters line in that division and that is bank one, and it does it very well. So you have one extreme hitters park and very good hitters park in one division. So now you have over 20% of your away games in a hitters park, again is it any wonder the other three look like very favorable pitchers parks? Am I saying they are not? No but they look a lot more extreme because of COors.

[SABR Matt]

blatantly neutral? Wrigley is not blatantly neutral. Wrigley Field for years and years and years was a hitters park. then in the 90's other parks come on line and then everybody is saying how the wind makes it neutral. Huh? The wind didn't exist in 1978? It's still a hitters park just not a Coors Field.


You can rest your freakin case but how is Great American Ballpark a known pitchers park? How is Turner Field a known pitchers park? The GAp has been open 3 years It had one neutral, one pitchers, and one hitters. Yet somehow you think you can rest your freakin case on that one? Turner Field has been open for 9 years and had two seasons in which its park factor was favoring the pitcher yet somehow you can rest your freakin case on it. These two parks you think are locks because they favor your view yet a park like citizen bank is too early to tell. PNC park has been around for 5 years and in that time it has hovered around league neutral yet you somewho see fit to declare it a pitchers park. Yet Wrigley looks similar to PNC the last 5 years but has a higher high is not a hitters park.

Again though the NL bottom is going to look a lot worse then the AL bottom because of Coors Field. I would love to see what the park factors for the NL would look like in the last 10 years without Coors Field.

Again...not true in either case.

One thing you should know...the park factors Randy and I are using are LEAGUE NEUTRAL...they don't use league-only statistics to verify, so Coors Field has no impact on how the other NL parks rate aside from its total impact on baseball as a whole.

GAB was one of the strongest pitchers' parks in baseball in 2004...I admit there's time for that to shift...as of right now, I don't see any reason to belief otherwise.

You refer to standard park factors to make your case about parks being "neutral"...but that's PART OF THE PROBLEM...with so many pitchers parks in the NL, NL-only park factors working on the assumption that the league by itself will average out to neutral are severely biased at present...this is a crucial thing to get across to anyone reading these posts...the league is NOT NECESSARILY park neutral...until people recognize this...we're going to have problems rating the parks in each league.

One Coors Field cannot erase 6 to 8 pitchers parks in the NL compared to just ONE in the AL.

Look I know it's early and we haven't done the legwork to go through the peer review process yet, because we're still working on additional methods for improving our accuracy, but I remain strongly convinced that we need to move away from multiplicative park factors that use league data only and are based on percent change in the RS at each park and move toward factors that define runs added or subtracted PER GAME...games or outs are the unit of time you play in each park...the park's net impact on scoring does not change with the number of runs you score in that park...it changes with how long you play there.

[SABR Matt]

Who plays in the NL west?

Colorado
LA
SanFran
San Diego
Arizona.

The majors now use a unbalanced schedule, so now these teams play a Coors field heavy schedule. Guess What? surpirse 3 of the 4 other teams play in a park that looks like very favorable pitchers parks. Gee is it any wonder when they play over 10% of their away games at Coors Field? Only one other stadium manages to hold its head up above the hitters line in that division and that is bank one, and it does it very well. So you have one extreme hitters park and very good hitters park in one division. So now you have over 20% of your away games in a hitters park, again is it any wonder the other three look like very favorable pitchers parks? Am I saying they are not? No but they look a lot more extreme because of COors.

Actually we rate San Francisco as a slight pitchers park...not an extreme pitchers park...and LA has moved more toward the cneter in recent years as well although it's still on the same level as Safeco Field.

Perhaps I haven't done my job attempting to explain the mathematical mechanics behind the FSAA, but the error you're concerned about in this post should not be a factor because when we calculate park factors for any one park we weight performances in those parks by games played for every team whoi played there. Coors Field shouldn't be having a downward impact on the other parks in the NL West because we don't compare home to away...we compare home to expected home scoring and the away data goes into measuring the other parks.

I'm gonna have to take some time and try to explain again how the math works in the FSAA to hopefully clear up why your concerns in the last several posts don't apply to the approach I'm using now...we're operating in a completely different paradigm than traditional park factors.

[SABR Matt]

Matt - I think you should find this link relavant. I don't agree with his analysis of home run hitting for reasons that I give in a comment on his site but his use of a mathematical analysis to evaluate a non guassian distribution that is highly skewed in a baseball context should interest you. http://www.arthurdevany.com. You want to go to his research paper on why home run hitting hasn't changed.

Your link doesn't seem to be working...the operation times out...not sure what's up with that.

I'm interested to see what this guy says though. :) And thanks in advance for the reference.

[misterdirt]

How odd. I just clicked on it as it appears in my post and it worked fine for me. I checked the address and its correct. If it doesn't work by clicking on the link in the post it should work by hand typing the address.

[SABR Matt]

Must have been a temporary server glitch...it's working now.

[SABR Matt]

Yeah...thyis is really fascinating...going to take me a while to read through it...but I'm impressed already.

[SABR Matt]

Well that was a really interesting read...thanks for supplying it...certainly changes the way I look at the home run distribution with time...

However...the Run Scoring distribution isn't the kind of extreme skewed "wild" distribution talked about here...normal analysis is probably still not appropriate since the RS distribution has a left bound and no right bound (thus causing the skew), but the HR distribution has skew of ~3 and kurtosis of ~13...the RS distribution has skew of ~1 and kurtosis of ~4. Big difference.

That's to be expected since scoring a run is a lot easier than homering...but...I'm not quite sure what to do with this now.

[Ubiquitous]

I'm not really concerned with Park Factors, nor was my intent to get in a debate about them. I think a lot of system wide general stats are flawed when looking at players. Its why I don't like stolen base metrics, ERA+, and park factors.

What I was talking about and interested debating is your belief that because of the DH the AL employs some players for their defense or a more defense oriented positional player. I disagree with that. I think the average positional player is about the same and both leagues and if their is a difference it probably can be explain by the top 1% of the leagues changing the league average and coors field as well. You have 8 starting spots in Coors field that get a pronounced boost to their numbers. You have Barry Bonds, Mark McGwire, Albert Pujols, and Sammy Sosa in the NL. Now you could argue that these guys are the big boppers of the majors because they play in NL parks but on the other hand these guys would be the big boppers of the AL as well and they would still be crushing the ball and still be altering the average.

[SABR Matt]

It is possible that the power hitters of the NL had an influence on the overal NL averages...but it would be a very small influence. 4 players out of 300 with regular playing time over that time span...I just don't think you can explain an entire league by four great players. And I think you vastly overstate the overall impact of Coors Field and make the assumption that the parks in both leagues will be about the same (excluding Coors Field) and therefore you can adjust the NL numbers down...I consider that a horrible misread of the data...park factors are central to my argument that the AL is a weaker hitting league if you exclude the DHs and pitchers...well not park factors...park adjustments...I've attempted to give some explanation as to how I arrived at the park factors I did...how I came to the conclusion that the NL is overall largely pitcher friendly and the AL is largely hitter friendly, but all I can really do is give you the logic that led to the answers...

[SABR Matt]

I must also admit to presently having not one iota of a clue as to how to fit a Levy stable distribution curve to a data set...I've read up on the subject and the only thing I've been able to say is "huh??"

[Ubiquitous]

It isn't just 4 hitters, but a lot of them. Is it possible that its the parks? Sure, but these guys were power hitters before they came to these parks. moved to the NL. Then you have the coors players, Larry Walker, Todd Helton, and a bunch of scrubs with inflated numbers (dante bichette, vinny castilla). With the last wave of free agency some of the sluggers have gone over to the AL though. Players like Sheffield and Vlad.

The NL had the greatest power hitting first basemen
Second Basemen
Third Basemen
Left Fielder
Center Fielder
Right Fielder
Catcher.

The AL had the SS, until ARod moved to third, I'll give them the best power hitting third basemen now, and they also keep the SS as well.

Is it because of the parks or the players? Well when I look at Mark McGwire, or Jeff Bagwell, or Albert Pujols, or Jim Thome I don't see a guy that needs a small park to put up huge numbers. Mike Piazza does not need to play in Coors field to crush some homers.

So for me I think the top 1% of the NL did or was affecting the average to a greater degree then the AL's top 1%.

As for Coors, year 2000
The NL in games not played at Coors field: .262/.339/.425
Coors Field: .318/..386/.525
The NL with Coors in the mix: .266/.342/.432.

Coors Field all by itself raised batting average 4 points, OBP 3 points, and SLG 7 points. So yes I do think Coors field does change the NL numbers. Its late but perhaps tomorrow I'll look at the top 1% and see how they changed the league average.

[SABR Matt]

It's things like that that continue to mystify me...you can't just look at the numbers players put up in the NL without Coors Field...then look at the total numbers and go "aha...I therefore have an excuse to slash all NL league totals to account for Coords Field. If you're going to adjust for one park...YOU MUST ADJUST FOR THEM ALL.

On the one hand, you argue that the NL had the better hitters...(at least in the top 1%)...and on the other...you claim the hitters were the same in both league and that I'm imagining the strong league/park bias. NO!! It doesn't work that way...you can't wave you hand and pretend one argument you make doesn't directly contradict the other.

If the NL has the best sluggers at most of the key positions, (and I do believe that it does)...that represents MORE TALENT IN THE NL amongst non-pitchers at the plate. And if you insist on adjusting for one park without adjusting for the others, then I see no point in continuing this conversation.

[Ubiquitous]

No I'm not saying you must adjust for one park and one park only. What I am saying is that Coors is a huge outlier. You say it isn't. Coors field effects the average like no other park out there. Thats all I'm saying.

Nor am I saying that one league is stronger then the other. I am merely disputing your view that the AL because of the DH employ more defensive minded players as a result. I think thats wrong. I don't think that a GM is going to put Rey Ordonez in the lineup because they have David Ortiz at DH.
The NL did that for awhile when they started the DH and they quickly realized that they needed to maximize their runs when using the DH.

As for talent. The top tier is no indication of league quality. Its the average and bottom that indicates league quality. The top tier is rare and extreme. Its sample size is small and very easy to alter thier ratio of AL:NL. If the NL has the top 5% of the baseball, that doesn't mean the average NL player is better then the average AL player.

For instance in 2004 if we remove the top 25 hitters in each league the rest come out to:
AL: .263/.330/.415
NL: .263/.329/.414

With the top 25 in
AL: .271/.338/.434
NL: .269/.341/.437

I should mention that I am only looking at positional players in both leagues. The top 25 hittes of the NL are impacting their league average much more then the top 25 in the AL.

[SABR Matt]

I don't see Coors as having the kind of overwhelming impact on the NL that you do...it's obviously one of the best hitters parks in major league history, but it's being balanced by one of the largest concentrations of (relative to the overall park environment of the modern era) pitching favored parks ever assembled in one league...and no...that's not an artifact of Coors messing with the away splits of other parks since (at least theoretically), all parks should behave independently and be relative only to an expectation to allow run scoring...not to each other as in the home vs. away paradigm of traditional park factors.

Just stop and think about this for a moment...

Even if I'm conservative and keep GAB (which just had a hiter friendly season it appears so it may be oscillating) and SBC out of it for now...

AL

Comisky II +
Skydome +
TBIA +
Fenway +
HHH Metrodome +
Safeco -
maaaybe Oakland though that's iffy -

NL

Coors +
Bank One +
Citizen's Bank Park +
maaaybe Minute Maid +
Shea Stadium -
Pro Player -
RFK -
Turner Field -
Dodger Stadium -
Petco Park -
PNC Park -

It doesn't take an einstein to see there's a difference in how the leagues are balanced...don't look at the standard park factor data...just use your common sense and think about how those parks are laid out...the climates they're in...and their overall reputation. Common park factors have been much discussed and most in the sabermetric field feel they are far too fickle and subject to error to be used with any confidence...just rely on your instincts.

[Ubiquitous]

How much does Coors field effect NL numbers? A lot. You say it is not enough to compensate for the all the pitchers parks. I believe it might very well be able to do that.

for instance looking at the 2000 numbers again, Coors field boost the NL numbers in average by 1.5%. The 5 parks that depress batting average effect the average by a total of 2%. Coors field compensates for 4 NL parks concerning batting average.

Coors field inflates OBP by .9%, three parks depress OBP and they do it by a total of .9%. So in terms of OBP Coors field absorbs all pitchers parks depressions.

Coors Field inflates SLG by 1.6%. 7 parks depress scoring by 3%. Coors field can gobble up 5 of those stadiums. Or it can gobble up the two biggest depressors and a smaller depressor.

Coors field is such a hitters park that it doesn't just cancel out 1 pitchers park, it doesn't just cancel out two pitchers parks. It cancels out around 4 pitchers parks. That something that no other park can claim.

Coors isn't just a "+", it's a "+++++"


On a sidenote I'm not really sure what you mean in your first paragraph. You mention expectation of scoring. I'm not sure how you come up with expected runs, but I have a feeling it has to do with league averages. Which of course has Coors field in it right?

[SABR Matt]

Not quite Ubiquitus.

The entire FSAA Matrix is based on the idea that run scoring occurs due to a combination of factors.

First - The all time scoring average. We've played so many games now that we have a good idea how many runs get scored on average by each side in a baseball game.

Second - The League Environment. It is Randy and my belief that not all of the difference between the all time average and the league average RS/G/Side is due to league factors. We believe there are differences in the parks, and the players in each league that account for some of that variation. The FSAA defines these things by looking at beta-distributed "Certainties"...the idea is that we can be increasingly certain a statistical effect exists with more data in the sample. A park effect created by playing three games at that park registers as a minor influence in the grand scheme of things because there's not enough information to be certain that effect is real and not the result of random variation. The larger the sample...the larger the certainty.

In the modern era, leagues rarely account for more than about +/- 0.3-0.5 runs per game per side above the all time scoring average.

Third - The intrinsic park adjustments. We also strongly believe that leagues don't necessarily have a net neutral park factor. Some league have better parks for hitters than others. Some of the differential between all time scoring average and league scoring average is often explained by the parks in that league.

Fourth - Team-specific reactions to parks. Not all of the variation in park run scoring averages is necessarily the result of the park itself...sometimes teams (especially home teams) can have a significant influence on how the park appears to play. We define that as a team skill, not a reflection on the park.

The strengths of team offenses and defenses. Remaining variations in run scoring are seen as the responsibility of the players making up the team units that do the scoring. For instance, in the 70s and 80s there was a glut of great fielders and good pitchers that helped keep the run scoring down even though the ball was not significantly different than the ball in the mid-90s (research has been done on this and REJECTED the notion that the mid-90s ball was significantly juiced...it WASN'T...we're just living in an era where the offensive players happen to be better than the pitchers...these things go in cycles). Variations in league run scoring cannot be entirely explained by variations in the conditions on the field...the strikezone has gotten smaller as have the parks, but this does not account for all of the run scoring increases of the modern game...

It's a serious mistake in just about every sabermetric analysis I've seen to base calculations entirely around league averages for this reason...we believe those averages are at least partially created by the balance of power among the players. We think the pitching and defense really was better in the late 60s through the mid 80s. We think the hitters really were better in the 20s/30s and 90s/00s.

These factors are all balanced through the laws of probability to determine the most likely arrangement of run scoring attribution (who/what was responsible for how much). Coors field should not have an impact on any one park's expectation to allow runs...at least not a significant impact...81 games is a small sample compared to the other 183,000 games against which it's being compared in any one season.

As to specifically how much of an impact Coors Field has...you're STILL making a tragic mistake...you're say Coors Field balances 4 other clubs in terms of the impact it has on the league's total production rate...the problem is, you're still looking at averages. A pitcher's park in a league with a lot of pitchers' parks isn't going to have as much of an impact on the overall league batting line as a pitcher's park in a league with a lot of hitter's parks (see: Field, Safeco...a pitcher's park, but the severity of said pitcher friendliness has been overstated because it's being compared to the rest of the AL) or a hitter's park in a league full of pitcher's parks (see: Field, Coors)

That's not to say Coors isn't a tremendous park for hitters...but it's only about 30% more hitter friendly than park #2 in baseball (TBIA).

[jalbright]

OK, Matt, I'm going to resurrect this thread and toss in links to two other threads that gave serious consideration to league quality adjustments:
http://baseball-fever.com/showthread.php?t=42166 from post 15 onward, with some other issues interspersed) and http://baseball-fever.com/showthread.php?t=42748 from post 24 or so. Have you continued with this idea of skew (which, if I understand it correctly, is essentially a measure of competitive balance, and in the Collins/Morgan thread, I noted the problem I have with using competitive balance as a stand-in for league quality).

We hit on some issues I have with the Schell standard deviation approach in the John Ward thread, but I'll add another point shortly.

Anyway, I think it is useful to collect these ideas on the topic for future consideration.

Also, Matt, does this thread contain the info on the paper about measuring the tail of the distribution curve you referred to in the Ward thread?

Jim Albright

[SABR Matt]

No...that came up in a thread discussing the hotly debated "increase" in the HR frequencies in the last ten years here in the sabermetrics forum.

I'm trying to locate it now.

[SABR Matt]

http://www.baseball-fever.com/showthread.php?t=35944

Second post...thanks to Ubiq for that link.

[jalbright]

I can propose another controlled experiment which I think shows the Schell approach is a flawed measure--even if you eliminated distortions by top players by using the middle quartiles. Once again, we start with a sim of a season. Eliminate the players who made the post season all-star teams (like TSN's) and replace them with players who are as close as possible to the mean for their given positions using the same number of outs or innings pitched. The league quality would dip--but the standard deviations of the league should get closer, which supposedly means we should interpret the results as a "better" league when in fact it is not.

I know I'm harping on these hypothetical controlled experiments, but a) I can force the question of league quality such that I know what the result should be, and b) if the measure can't successfully deal with such controlled tests of its ability to measure league quality, I have to doubt whether it is truly a good measure. That's not to say there isn't some correlation between standard deviations (which is, in essence, measuring standardization) or competitive balance and league quality. What I am saying is that it seems to me the correlations between standardization or competitive balance on one hand and league quality appear to me to be too weak to serve as an adequate stand-in.

The issue of standardization is one I think Bill James essentially addresses in his critique of Gould's theories in this regard in Whatever Happened to the Hall of Fame's Timeline chapter. I think to the extent baseball people have intelligently molded the evolution of ballplayers, there probably has been some progress due to the intervention of that intelligence. However, history is full of examples of far less than intelligent actions of the powers that be in baseball, so it's tough to see how much credit to give on that score.

I'm leaning toward the thought that we're going to have one hell of a time getting an adequate measure of league quality unless we resolve one of the two following issues:
1) a reasonable measure of the impact of aging; or
2) a reasonable estimate of replacement value.

The first would allow us to filter out a critical pollution of data on year to year measures of quality of play, as the players who are the basis for the analysis are always older in the more recent measure. The second might serve as a decent direct measure of the quality of play--but if it does not, it would serve as a reasonable floor for a proper analysis of talent (the tail of a normal distribution curve). I would think that with a proper anchor point for the floor level of the tail, mathematical tools for a reasonable measure of the average quality of play in a league should exist.

Presently, though, I tend to think we're either using numbers without any grounding in reality or are trying to use inadequate substitute measures of league quality. The main reason I describe these substitute measures as inadequate is 1) they can't seem to adequately deal with controlled experiments on known data, and/or 2) the measures include issues (such as distribution of talent) beyond league quality which seriously pollute the data and render our conclusions highly questionable at best.

Jim Albright

[SABR Matt]

Hey Jim.

I believe I *developed* a reasonable measure of replacement level.

Randy and I (some time ago) determined the pythagorean W% at which the average weakest teams peformed (and sub-teams...offenses and defenses) over time and found that replacement level rapidly increased through the 19th century and then tailed off to a gradual increase throughout the modern game.

This is entirely from memory but weak-team (10th percentile) pythagorean W%:

1876 -> .260ish
1900 -> .320ish
1925 -> .335
1950 -> .340
1975 -> .350
2000 -> .360

[Windy City Fan]

Unfortunately, I'm not nearly well versed enough in statistics to say if your methodology is correct, but the results are both interesting and reasonable. I've come to supsect that there was a fairly steady increase in the league depth up to the expansion era, and then expansion and new sources of talent combined to keep things at a relatively stable level in the late 70's and early 80's, after which league depth began to creep upwards again, but not nearly as quickly.

I'm not sure if the EDR is supposed to be representive of anything, but I wanted to see how well (if at all) it worked when combined with the relative stats OPS+.

Some people here, myself included feel Ted Williams has a case as a better hitter than Ruth, due to league quality adjustments (and the fact that Ruth was playing the OPS game when no one else was, but that isn't really accounted for in league quality issues). So I just wanted to do a quick test to see if the two stats combined would support this theory. Please tell me if such a quick and dirty combination of these two stats is not a valid use in your opinion.

Anyway, I just took Ruth's best 3 OPS+ seasons and Williams' best 3 seasons and multiplied them by the EDR for that league that year. Here's the results.

Babe Ruth
1920 AL 256 OPS+ .798 = 204
1921 AL 239 OPS+ .787 = 188
1923 AL 239 OPS+ .782 = 187

Ted Williams
1957 AL 253 OPS+ .795 = 201
1941 AL 235 OPS+ .870 = 204
1942 AL 217 OPS+ .854 = 185

Such a small test may mean nothing, but I wanted to get your take on how I combined the two stats and if the results of such a combination are valid in your opinion.

[digglahhh]

Jim,

I think the idea that we might have to throw out all our existing assumptions when dealing with extreme ends of talent distribution is not out of the realm of possibility.

I'm reminded of the scene in Good Will Hunting where the professors says to Will something about how there are so few people who can understand what they are doing and even fewer who can tell the differenence between the two of them.

Seriously, if you had only heard of Albert Pujols and his feats, but had never seen baseball played. But then you were given your first opportunity to see any kind of baseball played, and it was to watch Damian Miller take batting practice, you'd think, "That must be Pujols." Think about that.

I had a friend who was on the #1 ranked handball team in the country. He came down to my local park and beat the guy who some people thought was best player they'd ever seen, 21-2.

As you ascend the ranks of professional athletics, you approach the limits of the human body, the limits of physical coordination. That's why the mental aspect become so important.

League quality is an enigmatic notion to me. Think about how difficult it is to raise a 3.90 GPA to a 3.94. Does it really matter what the overall level of the rest of the class is? You are competing against the fundamental boundaries of achievement. The class will determine the curve, but you are so far ahead of it, that it doesn't really mean anything to you. If you are an A+ student you are going to trancend the "curve" whether the rest of the class is comprised of B students or C students, you are the "A" either way.

Now, say you composed a class of with a combined GPA of 3.94 and one of 3.9, what's the difference, pracitcally speaking? It seems to be a more aesthetic notion. If there is a student within either class with 4.0, is it particularly more impressive in either classroom?

I'm rambling a bit, and perhaps overindulging an analogy, but I think that the LQ adjustment is smaller than most think, especially relative to the players' own experiences. In fact, talent distribution is probably more of an issue than talent level in the earleir eras. Plus, I think it is prudent to keep in mind that these players compete against the limits of their human condition, especially the greats, who we speak mostly of.

The LQ may serve to confound more than clarify. So often, it is used in a general, rhetorical sense to prove the argument du jour.

[Ubiquitous]

Digg I think you are understating the League Quality impact a bit or at least comparing to something that doesn't apply.

A students GPA isn't based on other students, his grade isn't determined based on his performance against another student but against a standard system of work and questions. Do this and you get an A, do this and you get a B.

The problem with league quality is that everything is interacting. The standard numbers for excellence don't apply. You can't look at a league and say "well they averaged .295 as a group they must be good hitters. Therefore the league quality for hitters was good".

I think the true measurement in league quality isn't in the conventional stats or even in the standard deviations of conventional stats. Awhile back we had a topic about that, on whether SD shows league quality or if we the viewer use it to imply league quality. I was against SD for league quality. But anyway I think it isn't the stats that determine league quality but all the other things. Condition of the field, condition of the equipment, quality of umpires, how many umpires, how many bad decisions are made by managers, how many bad decisions made by runners, the kind of errors, the build of players, (and here is where stats could come into play) the amount of time mediocrity is allowed to play both in season and in years, so on and so on. I'm sure we could come up with a big list. I know Bill James has sort of this list already and I believe either Honus Wagner or DoubleX has posted it here somewhere.

[SABR Matt]

Jim,

I think the idea that we might have to throw out all our existing assumptions when dealing with extreme ends of talent distribution is not out of the realm of possibility.

I'm reminded of the scene in Good Will Hunting where the professors says to Will something about how there are so few people who can understand what they are doing and even fewer who can tell the differenence between the two of them.

Seriously, if you had only heard of Albert Pujols and his feats, but had never seen baseball played. But then you were given your first opportunity to see any kind of baseball played, and it was to watch Damian Miller take batting practice, you'd think, "That must be Pujols." Think about that.

I had a friend who was on the #1 ranked handball team in the country. He came down to my local park and beat the guy who some people thought was best player they'd ever seen, 21-2.

As you ascend the ranks of professional athletics, you approach the limits of the human body, the limits of physical coordination. That's why the mental aspect become so important.

League quality is an enigmatic notion to me. Think about how difficult it is to raise a 3.90 GPA to a 3.94. Does it really matter what the overall level of the rest of the class is? You are competing against the fundamental boundaries of achievement. The class will determine the curve, but you are so far ahead of it, that it doesn't really mean anything to you. If you are an A+ student you are going to trancend the "curve" whether the rest of the class is comprised of B students or C students, you are the "A" either way.

Now, say you composed a class of with a combined GPA of 3.94 and one of 3.9, what's the difference, pracitcally speaking? It seems to be a more aesthetic notion. If there is a student within either class with 4.0, is it particularly more impressive in either classroom?

I'm rambling a bit, and perhaps overindulging an analogy, but I think that the LQ adjustment is smaller than most think, especially relative to the players' own experiences. In fact, talent distribution is probably more of an issue than talent level in the earleir eras. Plus, I think it is prudent to keep in mind that these players compete against the limits of their human condition, especially the greats, who we speak mostly of.

The LQ may serve to confound more than clarify. So often, it is used in a general, rhetorical sense to prove the argument du jour.

I'm not saying I necessarily disagree with the assertion that the greats sort of "slip off the edge" when talking about league quality...but I don't think your example of the academic spectrum applies, mainly because, no matter what the class looks like as a group...any...and every...student is competing against himself and only himself. How you do in a class is more of less entirely dependent on your own application and ability and not dependent at all on the success of the other students.

Anyway...I tend to think LQ is a relatively minor tweak on the grand scale...I think the worst leagues have been above the quality of AAA today with a few exceptions...and the best leagues have only been slightly better than the game is today.

[digglahhh]

Ubiq,

The analogy isn't as off base as it appears. If you had a class of students who all performed very well, by normative distribution standards (like your league of .295 hitters) perhaps the material (league) was not challenging enough, or perrhaps they are, as a whole, a strong group of students (players).

My point is that, when you are taking the best students/players in the world, how important is the variation within that already incredibly selective group? Overall game advancements aside, which are moot in discussions of players relative to their own eras, how much random variation can one expect to occur within a group that is selected as the epitome of talent within a specific field at a given time, as opposed to a different time?

This is all rooted in the idea that when you have two players who dominated their leagues just about equally, one can pass the other based upon how easy that league was to dominate, so that is to assume that there is random variation in the talent level of the most talented players over time that is profound enough to impact the way the best of that already selective group are able to dominate the other most talented players. Any credible argument for that point, IMO, is going to have to focus on the leagues not being able to recruit the best available talent.

I tend to agree with your last paragraph. Peripheral conditions may provide advantages (opportunities) that certain types of players can take greater advantage of, and thusly overstate their abilities.

BTW,

The classroom dynamic is not just about a student against himself, it IS also about a student against others. For one, that's where the curve comes from. Also, submit a paper in an Ivy League institution and then in your local community college, see if the grade you are given has nothing to do with your competition...

[SABR Matt]

Most professors in the world today do NOT grade on a curve. Maybe they did in the dark ages when you went to college digg. :D

And it's not the competition that generates lower grades in an "ivy league" school (which BTW...the best academic progress is no longer made in ivy league schools but in upper class colleges in the public domain because there's more money there than there is at Brown or Princeton anymore...Penn State and UVA and William and Mary and UW and even the private schools which are not in the old ivory tower...those are the places where the competition is actually stiffest)...it's the EXPECTATIONS of the professors and the challenge of the curriculum. The profs don't look at the mix of student abilities and say "well this is a weak class so I won't grade too harshly"...they have certain expectations set by their departments on what that school wants to promote as a good education.

[Ubiquitous]

The grade you get in an Ivy school and JuCo has more to do with who is grading the paper and their expectations then your peers. But at the end of the day though an A is an A. The A at the JuCo college carries the same weight as the A in Harvard. Whereas in baseball a .300 average in DoubleA or in 1930 is not the same as a .300 in the majors or in say 1968. But again though you don't have a specific group of fellow students trying to keep your grade down while keeping their grade up. There is no balance their. In baseball somebody will always fail while somebody will always succeed. In school everybody can fail or everybody succeed. There is no balance in play.

how much random variation can one expect to occur within a group that is selected as the epitome of talent within a specific field at a given time, as opposed to a different time?

The simple answer? A ton of variation.
The very top is almost never consistent throughout history. We are talking about an extrememly small group of people who are the most fragile to changes. In that slight changes that keep just a few players away can alter their numbers immensely. While other groups (for instance the average) have a huge amount of people available and losing even a lot of them won't do much to the group. But in regards to baseball we have already seen the reserve clause, segregation, and poor talent management that the best in one era of MLB might not truly be the best in baseball of that era.

The best are not uniformly distributed and it is entirely possible that we do see clusters every now and then. Peaks and valleys and what have you.

[jalbright]

Hey Jim.

I believe I *developed* a reasonable measure of replacement level.

Randy and I (some time ago) determined the pythagorean W% at which the average weakest teams peformed (and sub-teams...offenses and defenses) over time and found that replacement level rapidly increased through the 19th century and then tailed off to a gradual increase throughout the modern game.

This is entirely from memory but weak-team (10th percentile) pythagorean W%:

1876 -> .260ish
1900 -> .320ish
1925 -> .335
1950 -> .340
1975 -> .350
2000 -> .360

Makes some sense, though using teams to get results for individuals may cause some problems. It would have to be worked out over time, and I'd suspect the 10% difference between 1900 and 2000 might be a tad high. It would seem to get past most of the theoretical objections, so then the next test is applying it to real life. The biggest problem might be dealing with it year to year (to compare 1943 to 1944 to 1945, etc.). As a lower limit for calculus formulas to deal with the tail of a normal distribution curve, it might work reasonably well.

One other thing that occurs to me in terms of league quality: hitting skill on an individual basis shouldn't be something that should improve much simply because it is based not so much on learned skills as reactions--and I'm unaware of evidence that human reaction times have improved to any significant degree in the past 130 years. Yeah, you can improve the pool of hitters you select from, and you can teach and encourage the value of taking a walk--but there's not that much more you can do.

Jim Albright

[SABR Matt]

Makes some sense, though using teams to get results for individuals may cause some problems. It would have to be worked out over time, and I'd suspect the 10% difference between 1900 and 2000 might be a tad high. It would seem to get past most of the theoretical objections, so then the next test is applying it to real life. The biggest problem might be dealing with it year to year (to compare 1943 to 1944 to 1945, etc.). As a lower limit for calculus formulas to deal with the tail of a normal distribution curve, it might work reasonably well.

One other thing that occurs to me in terms of league quality: hitting skill on an individual basis shouldn't be something that should improve much simply because it is based not so much on learned skills as reactions--and I'm unaware of evidence that human reaction times have improved to any significant degree in the past 130 years. Yeah, you can improve the pool of hitters you select from, and you can teach and encourage the value of taking a walk--but there's not that much more you can do.

Jim Albright

I don't think I understand your first paragraph...could you clarify what you mean by "should make a good base for calculus formulas for dealing with the tail of a normal distribution"...?

This could be very very important and I believe you may be onto something that could be used as an approach.

I think hitting skill varies more than you seem to believe...and it does so because hitters ARE getting faster and stronger today. Just as marathon runners are running faster and lasting longer...just as human scientists are getting smarter and smarter...humans are constantly driving to improve themselves and the human race and in baseball this expresses itself in better medicine, better conditioning, better knowledge of the game and the players, better scouting, better field conditions, better coaching, better minor league programs, better scouting, better living conditions, better hand-eye coordination, better pitch selection, better muscle mass buidling (for faster bat speed)...

[Ubiquitous]

hitting isn't just about reaction. Hitting is about seeing (which because of modern advancements is better), hitting is about power, hitting is about foot speed (getting to a base, getting to an extra base), hitting is about having the time and resources to practice, hitting is well a lot of things besides reaction and almost all of them have improved. And I'm willing to bet that as the human body gets into better condition gets into peak condition reaction improves as well.



I brought it up in this thread before but I don't think looking at teams answers the question in terms of player quality. Bill James does it as well but I think in the days of the reserve clause and now in the days of lopsided revenue streams it is entirely possible for the quality of players to be unevenly distributed in a league. Generally speaking in the old days around a quarter of the teams probably were of AAA or AAAA quality while in the middle you probably had major league teams and the top teams in terms of quality might very well be close to modern average teams.

[SABR Matt]

how are you defining "the old days" Ubiq...because if you're making that claim about anything other than the 1870s and 1880s then I'd respond with uproarious laughter. There's no way on earth that the rest of the teams in the majors ni the post-1900 period were that weak.

[Ubiquitous]

So when the St. Louis Browns were losing 100 games a year they were not even close to AAAA? Or when the Red Sox were losing 100 a year? Or the Senators? Or when Connie MAck did one of his purges? Or when the Braves were losing 100 a year? Or the Phillies?

[SABR Matt]

I think there were AAA-AAAA TEAMS...but not "1/4 of the league" as you suggested...and I don't think the good teams from back them were "maybe equal to an average team today"

[Ubiquitous]

1/4th of a league is two teams a league and 4 teams total for the entire major leagues.

As for great teams why not? Even the great teams back then didn't have significant depth. They couldn't stock their team with hitters at every position or even a 4 man rotation and had to settle for players that at times probably didn't belong on a roster.

[leecemark]

--I would say that the best minor league teams were frequently better than the worst major league ones for the first half of baseball history. If the Red Sox and Orioles had switched leagues in the early 20s I am fairly confident that the Orioles would have fared better than the Red Sox did. I wouldn't be surprised if they finished in the first division. I think the best PCL clubs were better than the worst MLB teams through the mid-30s or perhaps even up to WWII. The same would be true of the best Negro League teams.

[SABR Matt]

There were two good negro league teams...the rest were patsies.

Now the PCL I have argued for a long time WAS a major league back in the 20s/30s so I don't disagree that the best PCL teams would best the worst AL or NL teams.

But That's because I think the PCL was a 95+ league and MLB was a 95+ league...not because I imagine MLB to have been barely AAA calibar as you guys apparently do. I just find that utterly unsupportable.

[Ubiquitous]

So what about Connie MAck's teams makes you think that they were not AAAA? Or the Browns or the Phillies?

These teams would go stretches of years where they would 100 games or more and that is in a 154 game season. Top-tiered teams would for the most part absolutely dominate them. Their rosters filled with washed out vets, scrubeenies, and pimpled face kids. Yet I am supposed to believe that these teams were of similar quality to the top tiered teams of their day and of teams from nowadays?

If a minor league group could muster teams that were of equal value to the major leagues what does that say about the major leagues? What does that say about the major leagues and their ability to get the top talent?

[SABR Matt]

Um..hi...still not listening to me?

I have never said that Connie Macks' teams weren't AAAA...I just think it's ridiculous to claim that 1/4 of the game was AAA+ and that the top tier teams were no better than average teams of today (that last part I find especially ridiculous).

Remember, I believe MLB of the 1900-1950 period was probably roughly a 93-96 on the league difficulty scale compared to MLB of today...so I am not saying the leagues were the same as today...nor am I saying they didn't have laughing-stock teams.

But the top teams were the top teams because they cornered the market in great players...there's simply no way the top-tier teams were merely average by modern standards.

[Ubiquitous]

So teams a league were not AAAA? When the A's were bad all I have to do is find another team and presto a quarter of the league was bad. You had stretches where both the Braves and Phillies were bad.

Nor did I say that Very Good teams of then are no better then average teams of today. I said they might be close to average teams of today. That doesn't mean no better, that means around that level.

[Ubiquitous]

Remember, I believe MLB of the 1900-1950 period was probably roughly a 93-96 on the league difficulty scale compared to MLB of today...so I am not saying the leagues were the same as today...nor am I saying they didn't have laughing-stock teams.


If a league is around a 93-96 level of todays then league then that means that the laughing stock teams are somewhere in the low to mid 80's. A low to mid 80's team of today is a minor league team. Possibly a very good one but still a minor league team. If these bottom feeders played today they would have a record that was very spideresque and that might not even perform real well in todays minor leagues.

[Ubiquitous]

I just picked a one hundred loss team at random. It was the 1923 Boston Braves.

C: Batted .212/.258/.261 was 23 out of the majors at 27
1B: better known for his fielding, but this year was probably his best year with the bat in 5 years. Played only one more full season in the majors out of baseball two years after that.
2B: Good glove-no hit fielder who got moved to SS in later years. Typical year with the bat probably made him a dime a dozen in the minors.
SS: .251/.285/.309. Hitting was so bad that they turned him into a pitcher.
3B: Wasn't bad, not a great glove. died in a car crash that year
RF: Best hitter, batted .319/.383/.448
LF: lg avg OBP no pop, debut at 28, first year in the majors out of the majors in 5 years.
CF: Manned by two hitters. One out of the league after the next year the other a 36 year old who batted .264/.284/.352

Pitching:
Rube Marquard, sounds great but it was his last year as a starter and he would be out of the league in two years.
Joe Genewich: first year as starter, didn't do too bad
Jesse Barnes: comes over and does pretty good.
After that you have a whole collection of mediocre starters getting the ball.

This team had in its lineup a whole bunch of players that were not major leaguers. That if they played today would not be on a major league roster. The teams of that era still had to play this team and rack up stats against. To top it off they were not even the worst team in the league. The Phillies topped them by 4 games that year, and that team is a bit harder to look at because they played at the baker bowl.

[SABR Matt]

"Would not be on a major league roster"

So...apparently you're unaware that the 2005 Mariners ran a team of catchers who combined to hit for a sub .550 OPS.

Or that good-glove no-hit middle infielders populate juuuust about every major league roster in the game.

Or that a lot of young pitchers try and fail to stick in the big leagues every danged year.

Or that routinely...on all but the most elite of 2005 major league clubs, at least 5 of the 25 roster spots rotate between non-major leaguers.

[SABR Matt]

If the league is 93-96, the worst team is probably in the MID 80s...low-AAA calibar.

And the best team is far far FAAAAR from the 100 level you asserted (major league average)...probably more like 105 or 108.

Do you realize how bizarre it sounds when you claim the best major league teams were barely average compared to today and are apparently blind to the constant struggle even today to populate every big league roster?

[leecemark]

--The payroll imbalances that have grown so large over the past 2 decades have created an underclass of teams as bad, or nearly so, as those of yesteryear. Is there any real difference between the situation of the KC Royals and the old KC A's? Were the Browns any more hopeless than the Expos?
--There was a good run where the talent pool was close to maximized and no team was doomed to long term haplessness. The 60s, 70s and 80s (pick which was the best) fit that mold. If we are using the current majors as the 100 baseline, then 1975 would probably be something like 110 IMO (although the best teams/players player today may be as good or better - you can have a great team in a good league or even in a bad one).

[jalbright]

don't think I understand your first paragraph...could you clarify what you mean by "should make a good base for calculus formulas for dealing with the tail of a normal distribution"...?

This could be very very important and I believe you may be onto something that could be used as an approach.


I don't remember my calculus well enough, but if you have the end of a curve that is bounded on one side (replacement value) and you know the distribution with standard deviations (so you can describe the curve) and you know no one's going to beat a .950 winning percentage, you ought to be able to figure out a lot about the rest of the curve. Given the slope of the curve, I'd think you'd need calculus to make it work. At that point, my mathematical skills go south.

Actually, though, I had another thought that might well work--while competitive balance, skew of run scores, standard deviation measures of talent (overall and middle quartiles), estimates of replacement value, and maybe one or two others I haven't thought of individually aren't good substitute measures for league quality, what if we tried to graph them all with 100% being the median for each and then trying to create a graph of the results which achieves the best fit for all the different measures?

Jim Albright

[leecemark]

----1970s (107)
----1980s/1960s (105 - for 60s slightly higher in NL and lower in AL, 106/104)
----1993-2006 MLB (100)
----1950s NL (99)
----1950s AL (97)
----1930s AL (95)
----Current Japanese Legues (95)
----1930s NL (94)
----1920s AL (94)
----1920s NL (93)
----1910s AL (93)
----1920s PCL/1910s NL (92)
----1900s AL/1890s NL (91)
----1900s NL/Current AAA/WWII ball (90)
----1876-89 NL (89)
---American Association (87)
---Federal League/National Association (86)
---Current AA (85)

---I can't say with any assurance that these numbers hold up, but I am relatively sure the order of league strength. I do believe that the best teams are enough above their leagues that they could compete in a league as much as 10 percent better. The best players from even the weakest leagues I think could star in even the best ones.

[SABR Matt]

Jim...

Not sure I understand what you mean...

What would you be including in this graph?

I got that one element would be the skew of the run scoring distribution (which tells you something about competitive balance)...

I think the second measure that needs to be included is some measure of "major leagueness" of the talent. One way you might about doing that is to measure the average lifespan of players in a league. in 1945 for instance, the average lifespan of a position player was probably way less than it was in, say...1955...because there was a horde of war-babies who spent a few years in a watered down league and then faded away.

If you take each player who had any PT in a league...and determined how many PA he had in the big leagues overall (or DIO for pitchers)...and found a weighted average player-lifespan...you might get some information about how "major leagueish" a league is.

The only problem with that idea is...it won't work for the last 20 years of the game's history because you have active players who've just started their careers...are clearly of major league calibar, but do not register as such as of yet.

Another idea I had was to take each player's win-scoring rate in each season by normalized PCA and set it aside...then calculate the win-scoring-rate for the rest of his career besides the active season...and take a difference...if you do this for just one player, you're introducing a wrong-headed bias for flukiness of career arcs and for age...but if you do it for EVERY player in a league...you might get information about whether the league created a lot of flukishly good seasons (which would lead you to conclude that bad players were performing too well because the league sucked).

[SABR Matt]

Basically...I think there are two main variables that make up league difficulty.

One variable I believe I have already measured. It's difficult to star in a league where there is little separation in the teams and therefore no one to beat up on. That's why the skew of the Gamescore distribution works in describing competitive balance.

The other variable is like a displacement of that competitive balance.

Think of competitive balance like a bell curve. Once you get all leagues on the same bell curve, you need to do something to assure that the center of that bell curve lines up with all of the others. Each league must have the same average talent for the skewness measure to work.

So we need a second variable that describes the average tendency for the players in the league to be of major league calibar...that gives us the center of each league's bell curve.

Once you have that...you can line up all of the leagues...correct for the dkew, and bam...you're on an equal playing field.

[Ubiquitous]

"Would not be on a major league roster"

So...apparently you're unaware that the 2005 Mariners ran a team of catchers who combined to hit for a sub .550 OPS.

Or that good-glove no-hit middle infielders populate juuuust about every major league roster in the game.

Or that a lot of young pitchers try and fail to stick in the big leagues every danged year.

Or that routinely...on all but the most elite of 2005 major league clubs, at least 5 of the 25 roster spots rotate between non-major leaguers.


That team and many like it were replacement level players, and everybody knew they would be. This wasn't we're playing the youngster because we think they will be good. Or we have to play this guy because our main guy got injured. This was I am going to stock the entire team with a bunch of AAAA players and lose a lot of games. You say that today a lot of teams about 5 of the 25 are not major leaguers. Okay fine well about 18 or so of the 23 in those days were not major leaguers on the bad teams. Decent teams probably had up to 10 players and better teams probably reached the 5 or so that you claim today are not major leaguers. And on top of that you are ignoring your own adjustments. If the league was already lower to begin with then that means that the minor leagues and minor league level players were also lower in quality. If these teams were 93-95 on average and the worst teams were in the 80's then the minor leaguers wer at best in the 70's. Whereas today these AAAA players are probably in the mid 80's. So even if somehow it remained constant that only 5 players a team for most teams were below quality throught the era the 5 from today are vastly better then the 5 of yesteryear. But it wasn't constant there were more players back then who were not major league quality.

[Ubiquitous]

Do you realize how bizarre it sounds when you claim the best major league teams were barely average compared to today and are apparently blind to the constant struggle even today to populate every big league roster?


Are you even reading what I wrote? This will be the third time I have written this. I have never said barely average I said around average. I said the top teams would be around average. Does that mean I said the very best teams of yesteryear are barely average? No. Obviously if we were to put all the best teams of back then on top of each other it would look like a pyramid and the very best teams would be the most elevated teams away from average.


Secondly you keep disputing the notion that these leagues had very very bad teams. Teams that were way below what we have today but somehow it seems you are disputing bad teams by pointing out a statement on the very best teams of that era. Why? It's two different issues only connected by the fact these so called best teams played against some minor league quality teams. It doesn't make much sense to me to dispute bad teams by saying "well your statement on better teams is wrong. Therefore somehow you are wrong on bad teams"

[Ubiquitous]

If the league is 93-96, the worst team is probably in the MID 80s...low-AAA calibar.



So what the heck are you disputing? In fact it looks like your own metrics are telling you that these teams are even worse then I stated.

[SABR Matt]

I'm disputing your ridiculous notion that if the lowest teams in a league are 85 and the league is 94...that somehow...the best teams in that league are only 100 or so...that clearly cannot be true. My metric isn't really telling me anything useful about the exact percetage the league's quality was...it's just telling me something about how "competitive" the league was.

I'm disputing your idea that the 1930s Yankees were about the same as say...today's Minnesota Twins. Absolute hogwash and you know it.

[Ubiquitous]

You're disputing a notion that only you have a notion of.

But you know what? Part of what made some of those teams great was that they were playing against minor league teams. That they were facing pitchers that should have been in the minor leagues and hitters that should have been there as well. It's an 8 team league, and all it takes is two teams, a quarter of the league, to be minor league caliber to significantly alter (heck even one teams alters things dramatically) to significantly alter the playing field. Take an average team and have them play a quarter of their schedule against a minor league team and you tell me what their record is going to be?

An average team that beats up on a quarter of a league for a 154 game schedule could be expected to win about 90 games and they would still be average. 35 wins against the minor leaguers, 55 wins against everyone else. Ok it would probably be a little lower since the 44 games would have been against low quality teams just not that low but the point still stands. An average team would win about 77 games a year if the league did not have minor league teams in its fold. By having minor league teams an average team improves its record by about 10 or more wins a year. Now what about a good team? What about a team that could expect to win about 90 wins a year? They could expect to win about 100 games a year simply by playing a quarter of their games against minor leaguers.

To me all these avg to good to great teams got about a 10 or more win bonus simply by playing minor league teams. That to me is a fact (I'm sure don't think it is a fact, it's a saying)that cannot be ignored. These so called good teams are in part good because they played teams that were not up to par.

[Ubiquitous]

Secondly and again I'm not saying that the best teams were 100. 100 is in your view league quality average. Now that does not mean that todays leagues are at 100. The league quality could be at 103 or 105 right now. Which means if I think the best teams of yesteryear were around today's average then the best of the best would be around 107, 108, 109. Not 99 or 98. Even if we lock it into 100 as in these leagues now are 100 then I'm saying that the best teams are 103, 104, 105.

You've got the best teams of then between 105 and 108. We are off by a small amount of % points. Yet because you say 106 and I say 104 it makes my view completely ridiculous? I don't get that.

[SABR Matt]

Uh...the best teams today get to beat up on minor league teams too...ever heard of KC or Tampa Bay or (a few years ago) Detroit?

[SABR Matt]

I think the idea may have gotten lost in this bizarre fracas between Ubiq and I, but what are the thoughts on the idea of measuring the major-leagueishness of a league by the lifespan of the players in the league and/or the tendency of the league to cause anomalously good or bad seasons by seasonal sabermetrics?

I am personally somewhat curious to see what the lifespan idea presents...because I think if Ubiq's argument that the 1920s/1930s period (and before) included a lot of lame-duck teams with non-major-league players...more so than today...then we need to be able to prove that there were more marginal players and therefore more player turnover.

[jalbright]

I'll try and tackle it tomorrow--I'm too fried to handle that tonight. I tried earlier (in the afternoon), but the post went into cyber-nowhere, and I didn't have the time then to recreate it.

Jim Albright

[SABR Matt]

I'm in the midst of attempting to determine the average lifespan of a player in each league...but there are a number of caveats.

1) I've realized it is necessary to force every player to have the same opportunity to receive plate appearances. Becasue teams play different numbers of games at different points in history, and because in different run scoring environments, PA occur at higher or lower frequencies per game, I've calculated for each player in each team/season something I'm calling his APA (Adjusted Plate Appearances)...based on the differences in his teams' PA getting rates, and game-playing amounts.

For most players in the post-modern era, the APA is very very close to his real-world PA.

For players in...say...1876, the APA will be WAY larger...on the order of about 2.5 times larger despite the higher run scoring (and thus PA getting) rates per game.

With that out of the way, I need to total the APA for each player's career (easy), attach the total to each seasonal record (also pretty easy), total the APA for each league (easy), and then find a weighted average player career APA for the league...weights based on each player's APA as a proportion of the league's APA (extremely complicated).

I'll let you guys know what I find. Let's just say I have strong doubts about this idea that there was massively more AAAA player movement in the big leagues in 1930 than there is today. Side-note...this may have the side-benefit of helping me distinguish between the league's depth for position players and for pitchers since it may be true that there is a deep pool of established major league calibar pitchers or hitters in a given league while the other group is in flux.

[SABR Matt]

OK...I *really* think I'm on to something here.

In this chart I'm going to show you..."Life" as a variable is the average Career APA of a player in that league...take this as a measure of the major-league-ish-ness of a league....it's actually measuring the frequency that a player who's career is short will get significant playing time in a league.
lgID yearID Life
AA 1882 2724.6 - WEAK
AA 1883 3531.6
AA 1884 3110.1 - THIRD LEAGUE
AA 1885 4202.2
AA 1886 4114.7
AA 1887 4438.7
AA 1888 4369.5
AA 1889 4252.7
AA 1890 2408.1 - THIRD LEAGUE
AA 1891 4228.9
AL 1901 4001.7 - EXPANSION YEAR
AL 1902 4888.8
AL 1903 4626.1
AL 1904 4778.2
AL 1905 4462.2
AL 1906 4355.9
AL 1907 4362.1 - TALENT STAGNATION
AL 1908 4349.0
AL 1909 4467.7
AL 1910 4473.4
AL 1911 4489.9
AL 1912 4591.7
AL 1913 4733.4
AL 1914 4754.2 - FEDERAL LEAGUE NO EFFECT
AL 1915 4913.4
AL 1916 5215.6
AL 1917 5148.0
AL 1918 5100.6
AL 1919 5317.1
AL 1920 5346.6
AL 1921 5532.1 - AL VERY STRONG
AL 1922 5529.5
AL 1923 5097.3
AL 1924 5281.2
AL 1925 4837.4
AL 1926 4941.5
AL 1927 4988.3
AL 1928 4706.9
AL 1929 4761.2
AL 1930 4514.4 - STAGNATION REPEATS
AL 1931 4657.0
AL 1932 4821.8
AL 1933 5217.6
AL 1934 5049.4
AL 1935 5035.8
AL 1936 5133.0
AL 1937 5004.9
AL 1938 5101.5
AL 1939 4876.8
AL 1940 5039.7
AL 1941 4820.8
AL 1942 4498.8 - WAR NADYR
AL 1943 3931.0 - WAR NADYR
AL 1944 3422.5 - WAR NADYR
AL 1945 3077.2 - WAR NADYR
AL 1946 4116.6
AL 1947 4396.0
AL 1948 4128.4
AL 1949 4294.6
AL 1950 4315.3
AL 1951 4391.0
AL 1952 4134.1
AL 1953 4232.8
AL 1954 4343.5
AL 1955 4472.2
AL 1956 4657.9
AL 1957 4565.9
AL 1958 4604.4
AL 1959 4632.8
AL 1960 4687.8 - AL JUST RECOVERING...
AL 1961 4274.9 - EXPANSION KILLS IT
AL 1962 4108.4 - EXPANSION HITS HARD
AL 1963 4170.3
AL 1964 4257.8
AL 1965 4388.4
AL 1966 4641.2 - LEAGUE FINALLY RECOVERS
AL 1967 4627.8
AL 1968 4716.6
AL 1969 4408.1 - EXPANSION NOT AS DEADLY...LATIN WAVE?
AL 1970 4505.1
AL 1971 4731.7
AL 1972 4754.3
AL 1973 5309.1
AL 1974 5443.8
AL 1975 5799.0
AL 1976 5796.0
AL 1977 5221.5 - MINOR EXPANSION BUMP
AL 1978 5293.9
AL 1979 5407.0
AL 1980 5146.4 - ??
AL 1981 5522.5
AL 1982 5690.0
AL 1983 5661.3
AL 1984 5720.2
AL 1985 5863.6 - PEAK STABLE PERIOD
AL 1986 5851.8
AL 1987 5676.7
AL 1988 5606.8
AL 1989 5447.8
AL 1990 5622.4
AL 1991 5737.8
AL 1992 5635.8
AL 1993 5500.2
AL 1994 5576.1
AL 1995 5500.4
AL 1996 5260.4
AL 1997 5114.6
AL 1998 5082.0
AL 1999 4669.5
AL 2000 4535.8
AL 2001 4168.1
AL 2002 3613.1
AL 2003 3124.4
AL 2004 2961.3
FL 1914 2116.5 - NON-MAJOR LEAGUE
FL 1915 2482.4 - NOT MUCH BETTER
NL 1876 2905.2 - EW
NL 1877 3834.1
NL 1878 4454.3
NL 1879 4608.6
NL 1880 5522.5
NL 1881 5704.0 - ODDLY VERY STRONG
NL 1882 5595.0 - AA HAS NO IMPACT
NL 1883 5614.1
NL 1884 5286.3 - UA MINOR DIP
NL 1885 5418.4
NL 1886 5414.0
NL 1887 5556.1
NL 1888 5609.6
NL 1889 6041.9 - VERY!...STRONG
NL 1890 4312.0 - OWCH!! PL KILLS LEAGUE
NL 1891 6052.7 - RIGHT BACK
NL 1892 5889.3
NL 1893 5783.2
NL 1894 5620.7
NL 1895 5514.4
NL 1896 5549.0
NL 1897 5450.7 - LEAGUE THINS OUT
NL 1898 5532.4
NL 1899 5382.2
NL 1900 6391.3 - 8 TEAMS FOR ALL PROS
NL 1901 5558.2 - AL RAID BEGINS
NL 1902 4618.6
NL 1903 4567.2
NL 1904 4403.4 - OW
NL 1905 4589.8
NL 1906 4540.1
NL 1907 4394.5 - OW AGAIN
NL 1908 4602.7
NL 1909 4269.3 - MAN!
NL 1910 4526.9
NL 1911 4471.9
NL 1912 4520.8
NL 1913 4746.4
NL 1914 4438.7 - FL HITS NL HARDER
NL 1915 4447.6
NL 1916 4829.3
NL 1917 4584.2 - WWI?
NL 1918 4549.4 - WWI?
NL 1919 4693.7
NL 1920 4687.0
NL 1921 4426.6 - NL MUCH WEAKER
NL 1922 4339.0
NL 1923 4570.9
NL 1924 4684.6
NL 1925 4376.4
NL 1926 4364.2
NL 1927 4353.1
NL 1928 4638.6
NL 1929 4685.1
NL 1930 4642.7 - LEAGUES EVEN OUT A BIT
NL 1931 4494.2 - BUT NOT FOR LONG
NL 1932 4616.3
NL 1933 4706.7
NL 1934 4711.5
NL 1935 4521.7
NL 1936 4391.9
NL 1937 4302.7
NL 1938 4272.1
NL 1939 4254.8
NL 1940 4139.7
NL 1941 4226.7
NL 1942 4398.6
NL 1943 4016.9 - WAR NADYR
NL 1944 3350.7 - WAR NADYR
NL 1945 3143.3 - WAR NADYR
NL 1946 4090.0
NL 1947 4282.7 - NL CLOSES ON AL
NL 1948 4303.7
NL 1949 4434.5
NL 1950 4405.9
NL 1951 4509.0 - NL STRONGER LG
NL 1952 4498.7
NL 1953 4623.4
NL 1954 4888.9
NL 1955 4970.1
NL 1956 5372.6 - NL MUCH STRONGER
NL 1957 5063.0
NL 1958 5242.5
NL 1959 5288.8
NL 1960 5326.0
NL 1961 5517.1 - NL NOT EFFECTED BY EXPANSION #!
NL 1962 5140.9 - MINOR EXPANSION BUMP
NL 1963 5526.1
NL 1964 5406.2
NL 1965 5573.9
NL 1966 5543.3
NL 1967 5505.6
NL 1968 5381.7
NL 1969 5016.0 - AGAIN MINOR BLIP
NL 1970 5092.8
NL 1971 5294.7
NL 1972 5436.3
NL 1973 5535.5
NL 1974 5589.8
NL 1975 5431.1 - AL REASSERTS CONTROL
NL 1976 5398.5
NL 1977 5468.2
NL 1978 5533.8
NL 1979 5479.0
NL 1980 5415.4
NL 1981 5450.3
NL 1982 5523.0 - NL PEAK MUCH WEAKER
NL 1983 5452.9
NL 1984 5127.7
NL 1985 5187.5
NL 1986 5085.2
NL 1987 5039.9
NL 1988 5052.3
NL 1989 5075.1
NL 1990 5204.4
NL 1991 5133.6
NL 1992 5045.8
NL 1993 4749.3
NL 1994 4801.5
NL 1995 4565.0
NL 1996 4777.7
NL 1997 4762.3
NL 1998 4372.1
NL 1999 4267.7
NL 2000 4024.9
NL 2001 3949.6
NL 2002 3762.8
NL 2003 3506.1
NL 2004 3007.9
PL 1890 5561.1
UA 1884 1535.3 - NON-MAJOR-LEAGUE...WOW!!

The decline that occurs in the last 20 years of both the AL and NL is just caused by the fact that more and more "active" players are in the league and therefore their careers aren't finished and therefore they have low APA career totals...ignore that for a moment and focus on the years I'm marked with comments.

I think this is a pretty good trace of the history of talent depth in baseball.

[Ubiquitous]

Uh...the best teams today get to beat up on minor league teams too...ever heard of KC or Tampa Bay or (a few years ago) Detroit?

Yes and by your own admission these "minor league" teams are of higher quality then the so-called "minor League" teams of yesteryear. Also you are not doing the math. KC plus TB=2. 14 teams in the AL, interleague games that equals out to about 12% of a teams games are against minor league teams (which I don't think they are), but of course that doesn't count unbalanced schedule.


There are 30 teams in the game now and it would be a struggle to find 4 teams a year that are minor league level. Two would be closer to likely and that too would be a struggle. But lets say there is 4 teams just like there was several decades ago. 4 teams back then meant 25% of the teams were minor league quality. 4 teams now is 13% of teams are minor league quality. In order for it to be equal then one would need to find 8 teams to be of minor league quality in today's game. that isn't going to happen.

To me you could maybe find one team that is of AAAA level every now and then in todays game. You are not going to find 4, and you won't find it to be a regular occurrence even getting that one team. For instance the Pirates have been bad for awhile but if you look at the lineup and rotation one would know that it isn't a minor league team. Tampa Bay after a couple years past expansion was not a minor league team. Part of there problem is that they have three heavy spenders in there division with 2 of the teams being generally the best in the league. The quality of players on the Tampa team does not approach the level of mediocrity that I showed in the Braves team of 1923 (I think it was). Even Seattle wasn't a minor league team with as bad of talent as the teams of yesteryear. The closest I can find would be the Detroit Tigers of 2002 and 2003. That was a team in which a lot of the guys you knew were just not good and were not going to be good.

[Ubiquitous]

Looking at the theory I'm not sure it would actually prove what you are looking to prove or disprove.

One of the unique things about baseball back then was how often mediocre players were allowed to amass signifcant playing time. Every so often there is a thread about players who played the longest and were bad. who had the worst career so on and so on. These lists are dominated by players of yesteryears. Usually up the middle guys who teams ran out there day in and day out because quite frankly there just wasn't the money or enough players out there to fill every spot with good players.

Skeeter Newsome was allowed to rack up 4000 PA and was dreadful with the bat.
Rabbit Warstler allowed 4500 PA and he couldn't hit a lick
Lou Finney was a first basemen/RF who couldn't hit and he got 5,000 PA
And this was me looking on just one bad A's team. The Braves team I looked at was stocked full of players who got way more playing time then they deserved.

[csh19792001]

So what about Connie MAck's teams makes you think that they were not AAAA? Or the Browns or the Phillies?

What does that say about the major leagues and their ability to get the top talent?

Not saying I disagree, but...

The Devil Rays have averaged 97 losses per year. That's a .401 lifetime winning percentage, and it's way, WAY behind every single other franchise still in existence- that goes back to 1876. That's even worse than most of the stop gap/syndicate 19th century franchises (the ones that lasted, I mean) that people point to as proof of how pathetic 19th century ball was. Other teams like the Rockies and Padres have been similarly lousy overall. And this is supposedly the strongest/most competitively balanced league in baseball history.

[Ubiquitous]

Another example you have the 1921 AL listed as very strong. Yet the 1921 Athletics was a dreadful team, and the White Sox had to replace most of there starting team. Remove Cicotte and Lefty and the pitching staff looks like Kerr-Faber and a bunch of minor leaguers. Look at there ERA of everybody except Faber-Kerr.

The hitters were not too bad but it was an extremely thin team, and some spots did show through. Namely third-base and the bench. They had a nice platoon in Center but with a thin bench they had to ride there starters even if they were not the best of hitters.

I wouldn't call the White Sox a AAA team nor a AAAA team but a bad MLB team just like I would call Tampa a bad MLB team but I would call the A's a AAAA team.

So here is a league in which you call it pretty strong yet it has AAAA team in the league. The A's were pretty darn close to minor league from about 1915 to 1921. Though I will say that you probably don't see another minor league level team until about the end of the decade in the AL, with probably a AAAA like team every year occurring in the 30's. Over in the NL is a different story in that I think for a very long time one did not have to look far for a minor league team.

We've talked about it before when we were looking at standard deviations and so forth. But I have always thought the AL propped itself up at the expense of the NL. Meaning that there was never enough white talent on hand to fill 16 rosters and the AL with the better funding, ballparks, and popularity shifted the balance in their favor and kept it in their favor until Jackie Robinson. Pre-WWII seasons were probably the height of white dominated AL quality, and even then they couldn't amass enough talent to keep some teams from being truly awful.

[Ubiquitous]

Not saying I disagree, but...

The Devil Rays have averaged 97 losses per year. That's a .401 lifetime winning percentage, and it's way, WAY behind every single other franchise still in existence- that goes back to 1876. That's even worse than most of the stop gap/syndicate 19th century franchises (the ones that lasted, I mean) that people point to as proof of how pathetic 19th century ball was. Other teams like the Rockies and Padres have been similarly lousy overall. And this is supposedly the strongest/most competitively balanced league in baseball history.

The DRays are an expansion team and playing in a division with the Red Sox and Yankees. They are going to pile up the losses no doubt about it. There were some years where I would bet they were AAAA team but I also bet there were some years where they looked like AAAA teams but were not. they were just victims of a tough unbalanced schedule. I don't think last year they were a AAAA team nor 2004 nor probably 2003. They were probably a AAAA team there first year, but I'm not really sure they were a AAAA team the next year or the year after that. Then you have 2001 and 2002 the low point in terms of losses for the Rays. Were they a AAAA team? To coin a new phrase, maybe a AAAAA team. I say that because when I look at there roster and look at those A's and Braves and Browns teams roster I do see a difference. Those teams were filled with players who had no future, where there was no optimism. While on the TB there are some players that do, not enough to be a major league team possibly but there are more players on the Rays with that shine then in those teams I mentioned.

[digglahhh]

What would happen if you took some of the all-time greats from different eras and removed their performances against the worst X% of competition.

Remove, say, the lowest 10% of hitters and pitchers. Then determine new league averages. Eliminate the ABs each hitter had against pitchers who fall below the 10th percentile and then extrapolate the hitters' #s back to a full season's worth by filling those ABs back up based upon their "new" rate stats against the "new average pitchers." How much would it change the stats.

I know this would be a lot of work, but what if you tried this with different greats from different eras.

I know splits are incomplete. Maybe you'd just have to eliminate certain teams from the league to simplify things.

I know this, regardless of the difference between the quality of a decent major league pitcher (ex. Brad Radke) and a bad pitcher (ex. Victor Zambrano) Manny Ramirez is way better than both of them. The greats make a lot of guys look like AAAA pitchers, that's why I'm not sure how much it even matters in the grand scheme of things.

[SABR Matt]

Uh...the PA research PROVES what you're saying about the AL making itself the dominant league at the expense of the NL...and WOW...the NL gets back into it when Jackie Robinson signs!

Yeah...bad players got PT some of the time back then...but there are notorious bad players in today's game too. Royce Clayton, Rey Ordonez, Adam Everett, Brad Ausmus...

And more importantly, those bad players who had long careers would be very rare compared to bad players who did NOT have long careers...you're talking about a handful...maybe a dozen or two bad plaeyrs who had 4000 PA...but that's basically league average for a career...not some huge number...and if the league was truly weak...you would see the turnover...the fact that this method picks up the war, picks up the expansion years beautifully, picks up the weaker leagues and the stronger leagues the way we expect...tells me I'm on to something even if it's not a "complete" answer unto itself.

I think if you combine this...this PA theory...with the Skew of the RS distribution in some way (which would account for horridly bad teams who are perennially bad because horridly bad teams are punching bags that give up loads of runs)...you will get an answer that makes sense.

[Ubiquitous]

What would happen if you took some of the all-time greats from different eras and removed their performances against the worst X% of competition.

Remove, say, the lowest 10% of hitters and pitchers. Then determine new league averages. Eliminate the ABs each hitter had against pitchers who fall below the 10th percentile and then extrapolate the hitters' #s back to a full season's worth by filling those ABs back up based upon their "new" rate stats against the "new average pitchers." How much would it change the stats.

I know this would be a lot of work, but what if you tried this with different greats from different eras.

I know splits are incomplete. Maybe you'd just have to eliminate certain teams from the league to simplify things.

I know this, regardless of the difference between the quality of a decent major league pitcher (ex. Brad Radke) and a bad pitcher (ex. Victor Zambrano) Manny Ramirez is way better than both of them. The greats make a lot of guys look like AAAA pitchers, that's why I'm not sure how much it even matters in the grand scheme of things.

Problem is that no matter the level of play there will always be a bottom ten percent. I could have a 500 man league with 50 of them being 10 year old grade schoolers or a league where the bottom 50 are Bobby Bonds clones. A percent of total is too static(?), its a fixed baseline when there should not be one. In reality what one would need to do is find the true player ability or talent and then removes those players that are below a certain talent level. Then run the test.

[Ubiquitous]

Uh...the PA research PROVES what you're saying about the AL making itself the dominant league at the expense of the NL...and WOW...the NL gets back into it when Jackie Robinson signs!

Yeah...bad players got PT some of the time back then...but there are notorious bad players in today's game too. Royce Clayton, Rey Ordonez, Adam Everett, Brad Ausmus...

And more importantly, those bad players who had long careers would be very rare compared to bad players who did NOT have long careers...you're talking about a handful...maybe a dozen or two bad plaeyrs who had 4000 PA...but that's basically league average for a career...not some huge number...and if the league was truly weak...you would see the turnover...the fact that this method picks up the war, picks up the expansion years beautifully, picks up the weaker leagues and the stronger leagues the way we expect...tells me I'm on to something even if it's not a "complete" answer unto itself.

I think if you combine this...this PA theory...with the Skew of the RS distribution in some way (which would account for horridly bad teams who are perennially bad because horridly bad teams are punching bags that give up loads of runs)...you will get an answer that makes sense.

Of course its going to pick up a war. Everybody gets drafted, fill in a bunch of rosters spots several years later they are all forced out and the mediocrity of before the war is now too old. Of course its going to pick up expansion. Two teams that must fill there lineups with the dregs of other teams minor leagues. both events are not like finding a needle in a haystack. You could look at virtually any stat or use virtually any method and you are going to find the expansion years and war years.

Again though you seem to be forgetting your math. It is not as simple as finding one player for each player I name. You have to find two players to each of my names to have the same impact as my one name. 16 teams with a 23 man roster or so vs. 30 teams with a 25 man roster.

Nor does it prove what I was talking about when talking about proving and disproving. I wasn't saying it couldn't prove what league was better in the same year or era but whether or not amount of playing time of short careers proved anything in terms of league quality vs. another eras league quality.

Saying that it shows what I said doesn't prove that works. Since what I said was that the AL was better then the NL because of unevenly balanced spread of players but even with that advantage the AL still couldn't fill the rosters and some teams with major league quality players.

I said the late 30's was probably the high point for the AL whites, it looks as though your chart would say it started in the mid 30's and ended at the end of the 30's. Well the AL of that time had some godawful teams. Namely the A's and Browns. So here is a league that was a "high quality" league yet at times a quarter of it was manned by AAAA and AAA players.

[digglahhh]

Problem is that no matter the level of play there will always be a bottom ten percent. I could have a 500 man league with 50 of them being 10 year old grade schoolers or a league where the bottom 50 are Bobby Bonds clones. A percent of total is too static(?), its a fixed baseline when there should not be one. In reality what one would need to do is find the true player ability or talent and then removes those players that are below a certain talent level. Then run the test.

Doesn't that kind of lead you back to SDs?

[Ubiquitous]

Standard deviations is just %'s dressed in fancy formulas. 66%, 95%, 99.9%. All SD does is compare a players stats to the average, and again its going to be a fixed amount. Its impossible to have more then 5% of the players fall outside of 2 standard deviations. You are still rigging it to a fix total amount. The real total is a floating amount that can rise up and downs with the ebbs and flow of the game.

[Ubiquitous]

What I mean in terms of finding true talent level is sort of like the students grading at school. Instead of courses its skills based on ability and performance. Then you go through the list and remove any player below a 75 rating regardless of how many they are or whatever number it is that resembles a mass majority of minor leaguers.

Meaning if we do this for all baseball players at all levels we could then yes use SD, or %. We could say that in AAA the majority of players have a 75 rating. In double AA a 68 rating so on and so on. Then remove any player at the major league level that is a 75 or below and probably even say 75 to 80 or whatever rating one feels is "AAAA". Leaving just players that are on the low end of major league status to the superstars. In otherwords the true major leaguers and not a collection of major leaguers and scrubs interacting with one another.

[SABR Matt]

I should clarify something Ubiq...

I didn't claim that my little PA study "proved" the high "sucky player" rate idea was wrong...you might still be right that there was a great lack of true talent in baseball back then...

I merely offered the study as some evidence that perhaps the difference is not as great as you seem to believe. I think perspective is playing tricks on you a bit because you've seen a lot of lists of the worst players to have long careers which are (a) based on incorrect or poorly analyzed data (they don't give enough weight to fielding skill...they don't fully adjust for the scoring contexts...etc) and (b) non-scientific.

Here is the Normalized PCA bottom-feeder list who had at least 4000 PA and scored the worst in terms of career Wins/(648 PA).
First Last Ps PA WinRate
Lenny Harris UT 4211 4.53
Ricky Gutierrez SS 4126 4.51
Bill Kuehne 3B 4427 4.5
Neifi Perez SS 4514 4.5
Mike Pagliarulo 3B 4317 4.48
Roger Metzger SS 4674 4.47
Bill Coughlin 3B 4232 4.47
Tim Foli SS 6573 4.46
Ivy Olson SS 6630 4.46
Johnny Rawlings 2B 4165 4.41
Dan Wilson C 4588 4.36
Bob Kennedy UT 5063 4.31
Ed Sprague 3B 4587 4.29
Don Mueller RF 4590 4.29
Leo Durocher SS 5821 4.27
Wally Gerber SS 5816 4.22
Rafael Ramirez SS 5887 4.21
Mickey Owen C 4072 4.2
Chris Gomez SS 4307 4.2
Jerry Morales UT 4984 4.2
Ed Brinkman SS 6638 4.2
Ozzie Guillen SS 7133 4.18
Joe Quinn 2B 7334 4.14
Jose Pagan SS 4032 4.13
Joe Dugan 3B 5879 4.12
Rich Dauer 2B 4218 4.12
Wilbert Robinson C 5419 4.05
Darrin Fletcher C 4269 4
Tommy Dowd UT 5956 3.99
Bones Ely SS 5559 3.97
Sandy Alomar Jr. C 4592 3.96
Deivi Cruz SS 4100 3.92
Skeeter Newsome SS 4087 3.92
Rabbit Warstler SS 4611 3.88
Ken Reitz 3B 5079 3.85
Jesus Alou RF 4577 3.78
Barry McCormick UT 4029 3.72
Frank O'Rourke UT 4606 3.71
Joe Girardi C 4535 3.59
Aurelio Rodriguez 3B 7078 3.43
Rick Cerone C 4504 3.37
Jack Boyle UT 4632 3.07
Pete Suder UT 5473 3.02
Dan Meyer UT 4027 2.99
Art Whitney 3B 4001 2.98
Malachi Kittridge C 4446 2.79
Brent Mayne C 4084 2.77
Doug Flynn 2B 4085 2.66
Tommy Thevenow SS 4484 2.57
Gary DiSarcina SS 4032 2.04

Now...I don't know what you see on this list...but I do *NOT* see some kind of deep historical bias toward the older seasons...I see a good spread throughout time..with the worst player with significant PT being Gary DiSarcina and his ABYSMAL 2.04 W/648 PA rate...truly galling how horrible he was...and his career indeed in 1998.

That's people's exhibit B (exhibit A was the PA study) that this idea that there were more long-tracked bad-career players in the olden days. It doesn't look that way to me.

In fact Aurelio Rodriguez and Ozzie Guillen have the two longest careers on the bottom feeders' bracket and wow...they're both post-modern players.

Mind you...those win figures include both offense and defense.

I think there have always been bad teams and horrible players in this game...there continue to be bad teams and horrible players today...players who keep on getting long careers for one reason or another. Such is life in the game. Owners and GMs and Managers make bad decisions every day.

I understand (and have always kept in mind) your point about there bieng more roster spots, and the fact that the holes in the teams haven't expanded much despite the team sizes and numbers increasing suggests the league has gotten stronger...but then, you know as well as I do that I have always believed today's game is stronger than it was in 1920...we're arguing over a matter of degrees. I don't think a reasonable case can be made that the dominant teams of that time period were substantially worse than the dominant teams of today...at least not to the levels you've been suggesting.

[jalbright]

OK...I *really* think I'm on to something here.

In this chart I'm going to show you..."Life" as a variable is the average Career APA of a player in that league...take this as a measure of the major-league-ish-ness of a league....it's actually measuring the frequency that a player who's career is short will get significant playing time in a league.
lgID yearID Life
AA 1882 2724.6 - WEAK
AA 1883 3531.6
AA 1884 3110.1 - THIRD LEAGUE
AA 1885 4202.2
AA 1886 4114.7
AA 1887 4438.7
AA 1888 4369.5
AA 1889 4252.7
AA 1890 2408.1 - THIRD LEAGUE
AA 1891 4228.9
AL 1901 4001.7 - EXPANSION YEAR
AL 1902 4888.8
AL 1903 4626.1
AL 1904 4778.2
AL 1905 4462.2
AL 1906 4355.9
AL 1907 4362.1 - TALENT STAGNATION
AL 1908 4349.0
AL 1909 4467.7
AL 1910 4473.4
AL 1911 4489.9
AL 1912 4591.7
AL 1913 4733.4
AL 1914 4754.2 - FEDERAL LEAGUE NO EFFECT
AL 1915 4913.4
AL 1916 5215.6
AL 1917 5148.0
AL 1918 5100.6
AL 1919 5317.1
AL 1920 5346.6
AL 1921 5532.1 - AL VERY STRONG
AL 1922 5529.5
AL 1923 5097.3
AL 1924 5281.2
AL 1925 4837.4
AL 1926 4941.5
AL 1927 4988.3
AL 1928 4706.9
AL 1929 4761.2
AL 1930 4514.4 - STAGNATION REPEATS
AL 1931 4657.0
AL 1932 4821.8
AL 1933 5217.6
AL 1934 5049.4
AL 1935 5035.8
AL 1936 5133.0
AL 1937 5004.9
AL 1938 5101.5
AL 1939 4876.8
AL 1940 5039.7
AL 1941 4820.8
AL 1942 4498.8 - WAR NADYR
AL 1943 3931.0 - WAR NADYR
AL 1944 3422.5 - WAR NADYR
AL 1945 3077.2 - WAR NADYR
AL 1946 4116.6
AL 1947 4396.0
AL 1948 4128.4
AL 1949 4294.6
AL 1950 4315.3
AL 1951 4391.0
AL 1952 4134.1
AL 1953 4232.8
AL 1954 4343.5
AL 1955 4472.2
AL 1956 4657.9
AL 1957 4565.9
AL 1958 4604.4
AL 1959 4632.8
AL 1960 4687.8 - AL JUST RECOVERING...
AL 1961 4274.9 - EXPANSION KILLS IT
AL 1962 4108.4 - EXPANSION HITS HARD
AL 1963 4170.3
AL 1964 4257.8
AL 1965 4388.4
AL 1966 4641.2 - LEAGUE FINALLY RECOVERS
AL 1967 4627.8
AL 1968 4716.6
AL 1969 4408.1 - EXPANSION NOT AS DEADLY...LATIN WAVE?
AL 1970 4505.1
AL 1971 4731.7
AL 1972 4754.3
AL 1973 5309.1
AL 1974 5443.8
AL 1975 5799.0
AL 1976 5796.0
AL 1977 5221.5 - MINOR EXPANSION BUMP
AL 1978 5293.9
AL 1979 5407.0
AL 1980 5146.4 - ??
AL 1981 5522.5
AL 1982 5690.0
AL 1983 5661.3
AL 1984 5720.2
AL 1985 5863.6 - PEAK STABLE PERIOD
AL 1986 5851.8
AL 1987 5676.7
AL 1988 5606.8
AL 1989 5447.8
AL 1990 5622.4
AL 1991 5737.8
AL 1992 5635.8
AL 1993 5500.2
AL 1994 5576.1
AL 1995 5500.4
AL 1996 5260.4
AL 1997 5114.6
AL 1998 5082.0
AL 1999 4669.5
AL 2000 4535.8
AL 2001 4168.1
AL 2002 3613.1
AL 2003 3124.4
AL 2004 2961.3
FL 1914 2116.5 - NON-MAJOR LEAGUE
FL 1915 2482.4 - NOT MUCH BETTER
NL 1876 2905.2 - EW
NL 1877 3834.1
NL 1878 4454.3
NL 1879 4608.6
NL 1880 5522.5
NL 1881 5704.0 - ODDLY VERY STRONG
NL 1882 5595.0 - AA HAS NO IMPACT
NL 1883 5614.1
NL 1884 5286.3 - UA MINOR DIP
NL 1885 5418.4
NL 1886 5414.0
NL 1887 5556.1
NL 1888 5609.6
NL 1889 6041.9 - VERY!...STRONG
NL 1890 4312.0 - OWCH!! PL KILLS LEAGUE
NL 1891 6052.7 - RIGHT BACK
NL 1892 5889.3
NL 1893 5783.2
NL 1894 5620.7
NL 1895 5514.4
NL 1896 5549.0
NL 1897 5450.7 - LEAGUE THINS OUT
NL 1898 5532.4
NL 1899 5382.2
NL 1900 6391.3 - 8 TEAMS FOR ALL PROS
NL 1901 5558.2 - AL RAID BEGINS
NL 1902 4618.6
NL 1903 4567.2
NL 1904 4403.4 - OW
NL 1905 4589.8
NL 1906 4540.1
NL 1907 4394.5 - OW AGAIN
NL 1908 4602.7
NL 1909 4269.3 - MAN!
NL 1910 4526.9
NL 1911 4471.9
NL 1912 4520.8
NL 1913 4746.4
NL 1914 4438.7 - FL HITS NL HARDER
NL 1915 4447.6
NL 1916 4829.3
NL 1917 4584.2 - WWI?
NL 1918 4549.4 - WWI?
NL 1919 4693.7
NL 1920 4687.0
NL 1921 4426.6 - NL MUCH WEAKER
NL 1922 4339.0
NL 1923 4570.9
NL 1924 4684.6
NL 1925 4376.4
NL 1926 4364.2
NL 1927 4353.1
NL 1928 4638.6
NL 1929 4685.1
NL 1930 4642.7 - LEAGUES EVEN OUT A BIT
NL 1931 4494.2 - BUT NOT FOR LONG
NL 1932 4616.3
NL 1933 4706.7
NL 1934 4711.5
NL 1935 4521.7
NL 1936 4391.9
NL 1937 4302.7
NL 1938 4272.1
NL 1939 4254.8
NL 1940 4139.7
NL 1941 4226.7
NL 1942 4398.6
NL 1943 4016.9 - WAR NADYR
NL 1944 3350.7 - WAR NADYR
NL 1945 3143.3 - WAR NADYR
NL 1946 4090.0
NL 1947 4282.7 - NL CLOSES ON AL
NL 1948 4303.7
NL 1949 4434.5
NL 1950 4405.9
NL 1951 4509.0 - NL STRONGER LG
NL 1952 4498.7
NL 1953 4623.4
NL 1954 4888.9
NL 1955 4970.1
NL 1956 5372.6 - NL MUCH STRONGER
NL 1957 5063.0
NL 1958 5242.5
NL 1959 5288.8
NL 1960 5326.0
NL 1961 5517.1 - NL NOT EFFECTED BY EXPANSION #!
NL 1962 5140.9 - MINOR EXPANSION BUMP
NL 1963 5526.1
NL 1964 5406.2
NL 1965 5573.9
NL 1966 5543.3
NL 1967 5505.6
NL 1968 5381.7
NL 1969 5016.0 - AGAIN MINOR BLIP
NL 1970 5092.8
NL 1971 5294.7
NL 1972 5436.3
NL 1973 5535.5
NL 1974 5589.8
NL 1975 5431.1 - AL REASSERTS CONTROL
NL 1976 5398.5
NL 1977 5468.2
NL 1978 5533.8
NL 1979 5479.0
NL 1980 5415.4
NL 1981 5450.3
NL 1982 5523.0 - NL PEAK MUCH WEAKER
NL 1983 5452.9
NL 1984 5127.7
NL 1985 5187.5
NL 1986 5085.2
NL 1987 5039.9
NL 1988 5052.3
NL 1989 5075.1
NL 1990 5204.4
NL 1991 5133.6
NL 1992 5045.8
NL 1993 4749.3
NL 1994 4801.5
NL 1995 4565.0
NL 1996 4777.7
NL 1997 4762.3
NL 1998 4372.1
NL 1999 4267.7
NL 2000 4024.9
NL 2001 3949.6
NL 2002 3762.8
NL 2003 3506.1
NL 2004 3007.9
PL 1890 5561.1
UA 1884 1535.3 - NON-MAJOR-LEAGUE...WOW!!

The decline that occurs in the last 20 years of both the AL and NL is just caused by the fact that more and more "active" players are in the league and therefore their careers aren't finished and therefore they have low APA career totals...ignore that for a moment and focus on the years I'm marked with comments.

I think this is a pretty good trace of the history of talent depth in baseball.
Matt,

Am I correct in understanding that the top players affect this average? It should, since the overall quality of a league is significantly affected by its top talent, and odd concentrations of talent do occur by simple chance (see AL 1B in the 1930's with Gehrig, Foxx and Greenberg).

Also, does expansion help increase this measure for years prior to the expansion? It seems that providing more jobs might help prolong careers, thereby increasing the measure of preexpansion quality.

Assuming that you have addressed the expansion issue and measured the "average" player including the top ones, can you then combine your measures of skew (competitive balance) and this one into one measure of league quality? The model I'd propose is that 100 is the median, 105 very near the top and 95 very near the bottom of post-1900 leagues for both measures. The reason for this is our shared assumption/conclusion that few leagues would fall to the level of a minor league to another, and using the notion the spread in quality between majors and AAA today is 10%. Once both measures are fitted into these guidelines, you could generate averages of the results.

The only remaining issue would be for more current leagues, but if you limited use of the measure to retired players, you'd limit your problems there as well. You'd still have difficulties with guys who retired in the past dozen years or so, but it would be a long way up from where we are.

Jim Albright

[SABR Matt]

On question A...great players impact this measure only in that they tend to get a lot of playing time...although Greenberg didn't...for example. It's just APA counts...not a measure of player performance. Some pretty mediocre players have gotten 12000 APA careers because they had other skills people found desirable so the correlation isn't 1 to 1. If a league had a bunch of HOFers playing at the same time there's a chance it could slightly increase the average career APA of a player in the league, but only if those HOFers didn't come at the expense of a lot of turnover in other roster spots.

On question 2, I intend today to look at the problem of combining this measure with the skewness measure as well as one other measure.

This gives you an idea of how fast players arrived and left a league...a lot of turnover is a bad sign for that league.

The skewness measure tells you how competitive the league was internally. If you don't have a lot of player turnover, but that's because you've got a stagnant talent pool and some pretty stupid GMs, you'd probably see inreasingly large gaps between the haves and have nots (see...the second half of the 1930s) and skewness will pick up on that.

There's also the matter of determining whether, while this turnover is occuring, the other players in the league are having anomalously good years. I want to look at the average delta rate for the league (delta rate = scoring rate for the season in question minus career scoring rate in years other than the season in question)...if a lot of bad players have anomalously good years...that's a sign of poor league quality as well.

I think if we assume a league that NEVER had a major leaguer persist outside of the seasons it existed, we would call that league A calibar (75 on the difficulty scale)...so the closer you are to zero in terms of Average career APA, the closer you should be to 75...and if we set the median of APA to 100, rating on the proper scale the APA seasons would definitely be doable. For skewness, the larger your skew value gets, the more your league is becoming like a softball league. Softball would rate as a 50 on the league difficulty scale...below NCAA (70) and the top divisions of high school baseball (60). The skew in the 1870s is hilariously extreme and makes me believe that they were playing, more or less college baseball back then but it rapidly accelerates down the skew graph to values approaching major league quality. If I set the median skew value to 100 and place the 1870s somewhere in the 60s...the rest should follow pretty quickly.

[SABR Matt]

In fact I just factored out the active season from my lifespan calculation. If we're rating the stability of the 1945 AL's talent pool, we shouldn't include 1945's PA in the appraisal of the players who make up that talent pool.

With that done...the range is now 1161 (UA 1884) to 5389 (1991 AL). One problem with that appraisal is that the DH has prolonged careers in the AL since its' inception...but once I do this same thing for pitchers, I should be able to cancel that out (the DH probably shortened the careers of some pitchers too).

[digglahhh]

What I mean in terms of finding true talent level is sort of like the students grading at school. Instead of courses its skills based on ability and performance. Then you go through the list and remove any player below a 75 rating regardless of how many they are or whatever number it is that resembles a mass majority of minor leaguers.

Meaning if we do this for all baseball players at all levels we could then yes use SD, or %. We could say that in AAA the majority of players have a 75 rating. In double AA a 68 rating so on and so on. Then remove any player at the major league level that is a 75 or below and probably even say 75 to 80 or whatever rating one feels is "AAAA". Leaving just players that are on the low end of major league status to the superstars. In otherwords the true major leaguers and not a collection of major leaguers and scrubs interacting with one another.

Yeah, but tests that are well prepared and given to a large body of takers are designed so that the distribution of scores falls in a shape resembling the bell curve, based on SDs.

Again, remember this is an attempt to magnify the what is the very tip of the bell curve as is.

The "68s" and "75s" are really more like the "99.968s" and "99.975s" when you take into account how selective this group already is.

Has it been proven that within any portion of the distribtion on a curve, that if you look at only those who fall in say, the 67th percentile, that the distribution within that percentile mirrors the distribution curve of the whole sample? I don't know. Somebody else here might though. But, I tend to doubt it.

SDs are more than glorified percentages, SDs measure the homogeniety of data within a sample. Low SDs should indicate a high level of competitiveness, which is not, in and of itself, an indicator of high LQ, but it is a pre-requisite thereof.

[jalbright]

Matt,

OK so far, but you didn't answer about how expansion might affect the APA for prior years. It would be nice to eliminate or reduce the impact of that issue if it exists--and even if it can't be eliminated or reduced, to recognize the issue if we're going to try and use the measure despite that fact. I'll also be interested in hearing about the third measure.

I expected that at least early in professional baseball, we'd have results way outside the range I was talking about. However, I would seriously question any measure which indicated that anything but the very worst major leagues since 1901 were minor leagues compared to anything but some of the very best leagues since then. "Since 1901" is a key caveat in that statment.

Jim Albright

[Ubiquitous]

Yeah, but tests that are well prepared and given to a large body of takers are designed so that the distribution of scores falls in a shape resembling the bell curve, based on SDs.

Again, remember this is an attempt to magnify the what is the very tip of the bell curve as is.

The "68s" and "75s" are really more like the "99.968s" and "99.975s" when you take into account how selective this group already is.

Has it been proven that within any portion of the distribtion on a curve, that if you look at only those who fall in say, the 67th percentile, that the distribution within that percentile mirrors the distribution curve of the whole sample? I don't know. Somebody else here might though. But, I tend to doubt it.

SDs are more than glorified percentages, SDs measure the homogeniety of data within a sample. Low SDs should indicate a high level of competitiveness, which is not, in and of itself, an indicator of high LQ, but it is a pre-requisite thereof.

I guess I'm not seeing how with SD we could elimnate minor league players. Perhaps it's because I don't sue them in everyday life. For instance lets say we have a league with 500 men. In that league we know 20 of them are not major league quality. How would SD find them? Then we have another 500 man league and in that league there are no minor leaguers. Not saying they are all good just that even the worst are simply bad major leaguers and not good minor leaguers or anything like that. How would SD not remove any players or I should say how would it show that they are all major leaguers.

Yes one could argue that the spread would be smaller so perhaps at a certain spread it shows no minor leaguers or a players score could show he is a minor leaguer, but couldn't there be some factors on the positive side and around the middle to help hide some of this. Meaning if there isn't a lot of greats or the greats do not look all that great wile the rest are concentrated around the middle and that middle is somewhat low to begin with wouldn't that disguise a minor leaguer? Granted this is a hypo that might never exist in real life, but are there other more realistic situations in which a minor leaguer would be considered a major leaguers or a major leaguer considered a minor leaguer?

[SABR Matt]

Agreed Jim...I would have great difficulty accepting that any league overall since 1901 was worse than a 90 or so...more like 92 or 93.

I guess I kind of understand what you're saying about this pre-expansion effect...I'm not sure I see how a few players getting their careers extended by the expansion game would really significantly alter the league's average APA. It might provide a little tweak up in the five or seven years leading up to expansions...I doubt it would amount to much. Do you have a suggestion as to how I might look for that effect?

New data...here's the league register in the same order, but now I've calculated delta-Ws for each player. Using his career scoring rate (minus the active season's PA and W) and comparing it to the actie season, we can see the number of wins the league added to the player pool overall:
Lg Yr WinsAdded
AA 1882 52.44
AA 1883 34.39
AA 1884 95.73
AA 1885 13.45
AA 1886 8.51
AA 1887 2.80
AA 1888 8.89
AA 1889 7.15
AA 1890 119.01
AA 1891 26.01
AL 1901 33.19
AL 1902 -5.56
AL 1903 6.38
AL 1904 4.41
AL 1905 18.96
AL 1906 9.93
AL 1907 -5.70
AL 1908 1.56
AL 1909 -7.13
AL 1910 -2.89
AL 1911 8.35
AL 1912 9.02
AL 1913 -2.69
AL 1914 8.80
AL 1915 9.67
AL 1916 -6.14
AL 1917 -8.52
AL 1918 5.48
AL 1919 -19.49
AL 1920 3.27
AL 1921 12.64
AL 1922 -6.75
AL 1923 11.29
AL 1924 -7.53
AL 1925 8.95
AL 1926 -8.26
AL 1927 -1.80
AL 1928 12.47
AL 1929 12.13
AL 1930 10.94
AL 1931 15.78
AL 1932 15.70
AL 1933 -7.39
AL 1934 -11.49
AL 1935 7.64
AL 1936 -6.94
AL 1937 -9.46
AL 1938 -11.32
AL 1939 -16.87
AL 1940 -26.83
AL 1941 -12.88
AL 1942 -8.15
AL 1943 8.32
AL 1944 51.37
AL 1945 80.20
AL 1946 -9.54
AL 1947 -10.94
AL 1948 4.96
AL 1949 10.61
AL 1950 3.07
AL 1951 -1.47
AL 1952 13.83
AL 1953 20.13
AL 1954 16.18
AL 1955 8.01
AL 1956 1.11
AL 1957 17.34
AL 1958 7.56
AL 1959 2.81
AL 1960 -1.95
AL 1961 23.62
AL 1962 36.30
AL 1963 33.30
AL 1964 18.51
AL 1965 22.96
AL 1966 7.45
AL 1967 0.42
AL 1968 -2.82
AL 1969 56.22
AL 1970 53.75
AL 1971 44.64
AL 1972 65.69
AL 1973 -35.29
AL 1974 -34.15
AL 1975 -65.53
AL 1976 -33.30
AL 1977 4.41
AL 1978 -19.88
AL 1979 -4.74
AL 1980 34.84
AL 1981 -3.80
AL 1982 -2.34
AL 1983 10.86
AL 1984 -17.94
AL 1985 -27.86
AL 1986 -20.95
AL 1987 -5.35
AL 1988 -9.07
AL 1989 -3.18
AL 1990 -33.60
AL 1991 -54.05
AL 1992 -53.45
AL 1993 -28.18
AL 1994 -35.86
AL 1995 -36.41
AL 1996 -46.65
AL 1997 -57.32
AL 1998 -56.92
AL 1999 -42.53
AL 2000 -34.63
AL 2001 -22.92
AL 2002 -5.64
AL 2003 -6.02
AL 2004 13.03
FL 1914 60.86
FL 1915 51.39
NL 1876 44.99
NL 1877 -1.68
NL 1878 -2.21
NL 1879 13.89
NL 1880 -12.81
NL 1881 -17.78
NL 1882 -9.41
NL 1883 -11.69
NL 1884 14.23
NL 1885 -1.59
NL 1886 -5.59
NL 1887 -15.50
NL 1888 -14.18
NL 1889 -21.32
NL 1890 37.19
NL 1891 -41.31
NL 1892 -47.78
NL 1893 -26.35
NL 1894 -15.10
NL 1895 1.77
NL 1896 -8.05
NL 1897 -4.05
NL 1898 -17.60
NL 1899 10.50
NL 1900 -68.70
NL 1901 -11.63
NL 1902 40.30
NL 1903 9.94
NL 1904 20.01
NL 1905 11.28
NL 1906 16.17
NL 1907 3.17
NL 1908 -19.64
NL 1909 13.36
NL 1910 3.39
NL 1911 10.59
NL 1912 -19.28
NL 1913 -25.86
NL 1914 -10.88
NL 1915 -1.09
NL 1916 -23.21
NL 1917 6.66
NL 1918 8.94
NL 1919 4.40
NL 1920 12.48
NL 1921 27.38
NL 1922 28.33
NL 1923 22.47
NL 1924 6.72
NL 1925 15.26
NL 1926 27.83
NL 1927 10.42
NL 1928 7.73
NL 1929 -10.78
NL 1930 3.65
NL 1931 7.87
NL 1932 -12.77
NL 1933 -16.69
NL 1934 -6.18
NL 1935 -17.99
NL 1936 -0.26
NL 1937 -8.25
NL 1938 -9.97
NL 1939 -10.55
NL 1940 -1.01
NL 1941 -0.48
NL 1942 -18.34
NL 1943 4.91
NL 1944 48.17
NL 1945 60.69
NL 1946 5.30
NL 1947 -6.58
NL 1948 -2.00
NL 1949 0.23
NL 1950 5.12
NL 1951 2.09
NL 1952 13.03
NL 1953 13.30
NL 1954 -2.22
NL 1955 0.08
NL 1956 -14.51
NL 1957 3.77
NL 1958 -3.94
NL 1959 -11.54
NL 1960 -1.69
NL 1961 -4.50
NL 1962 21.43
NL 1963 -3.63
NL 1964 3.69
NL 1965 -16.71
NL 1966 -5.81
NL 1967 -3.04
NL 1968 11.07
NL 1969 45.08
NL 1970 22.18
NL 1971 24.51
NL 1972 21.67
NL 1973 24.39
NL 1974 23.87
NL 1975 37.80
NL 1976 40.53
NL 1977 30.97
NL 1978 18.40
NL 1979 26.20
NL 1980 38.54
NL 1981 19.42
NL 1982 9.78
NL 1983 27.38
NL 1984 38.48
NL 1985 32.58
NL 1986 20.12
NL 1987 9.66
NL 1988 4.44
NL 1989 20.53
NL 1990 -0.69
NL 1991 3.77
NL 1992 15.55
NL 1993 56.99
NL 1994 23.97
NL 1995 58.88
NL 1996 50.54
NL 1997 28.75
NL 1998 52.61
NL 1999 37.08
NL 2000 37.66
NL 2001 6.99
NL 2002 -0.82
NL 2003 30.89
NL 2004 24.00
PL 1890 -17.28
UA 1884 115.50

Once again...we see the UA standing out as utterly ridiculous. Joining the UA on the chopping block of non-major leagues is the AA in 1890 which suffered from a massive run on its' main talent base to the PL, which incidentally appears to be pretty stable by comparison. Interesting point of note...the AL in the 1990s has been the tougher league to dominate (as measured by skew analysis) and it appears that is a result of increased league difficulty...it looks like a lot of the major league talent has sloshed from the NL to the AL in the last decade or so. Could help to explain the enormous HR totals in the NL of late.

I don't think any one of these three measures, on their own, picks out league difficulty accurately...but I am of the opinion that a sythesis of these three metrics will do a good job...at least a more objective and less arbitrary method than, say, James' timeline adjustment or the timeline adjustment used in WARP3.

One of the unique things about my wins-added analysis is...my wins are already normalized. The win figures I used were the normalized win figures and hence I've already factored out the differences in standard deviation that exist in different run scoring environments. What you're seeing is completely devoid of any biasing effects of the run scoring environment.

[jalbright]

I guess I kind of understand what you're saying about this pre-expansion effect...I'm not sure I see how a few players getting their careers extended by the expansion game would really significantly alter the league's average APA. It might provide a little tweak up in the five or seven years leading up to expansions...I doubt it would amount to much. Do you have a suggestion as to how I might look for that effect?

I think we'd expect the effect, if it exists, to arise from say the bottom third or quarter of players in terms of productivity (the others are probably good enough that expansion would have a minimal effect on them). If all of a sudden in the few years before an expansion that group has a spike in APA, I'd think that would tell us something about a) whether it exists, and b) the size of the effect if it does.

Jim Albright

[SABR Matt]

Of course...if that group is having a productivity spike, we'd expect to see a positive number appear in the league's productivity...my third element.

That hasn't really been the case.

[SABR Matt]

In fact the opposite has almost universally been true. Right before expansions, the leagues appeared to be more balanced and more difficult to perform in that right after...there is no spike in performance prior to an expansion...there is one DURING expansion though.

[digglahhh]

I think we'd expect the effect, if it exists, to arise from say the bottom third or quarter of players in terms of productivity (the others are probably good enough that expansion would have a minimal effect on them). If all of a sudden in the few years before an expansion that group has a spike in APA, I'd think that would tell us something about a) whether it exists, and b) the size of the effect if it does.

Jim Albright

Agreed.

Now, wouldn't that actually make it harder for the greats to separate themselves?

I tend to think under the assumption that the best players are actually impacted least by LQ. Furthermore, if the marginal players do better, that inflates the league average and quite possibly deflates relative stats.

Other than obvious cases that have clear historic reasons for particularly bad LQ, this may very well be more trouble than its worth.

[jalbright]

Agreed.

Now, wouldn't that actually make it harder for the greats to separate themselves?

I tend to think under the assumption that the best players are actually impacted least by LQ. Furthermore, if the marginal players do better, that inflates the league average and quite possibly deflates relative stats.

Other than obvious cases that have clear historic reasons for particularly bad LQ, this may very well be more trouble than its worth.

You may well be right about the last sentence, except that if we can explain why we've come up with a better mousetrap that shows far smaller changes than many suppose, we can hope to at least rein in these wild assumptions about league quality.

As for it being harder for the greats to separate themselves, it might limit it to some degree in this measure, but I'm not sure that the best players don't get at least as much a boost out of expansion as the marginal guys in terms of performance.

I think the thing that encourages me about this idea is that I can't come up with a controlled sim experiment beyond the serpentine versus team 1, team 2, etc draft which would have a serious effect on the measure--and that one effect would be limited to 1/3 of the measure at most. Maybe I will over the next few days, but at least it's a start.

Jim Albright

[SABR Matt]

Agreed.

Now, wouldn't that actually make it harder for the greats to separate themselves?

I tend to think under the assumption that the best players are actually impacted least by LQ. Furthermore, if the marginal players do better, that inflates the league average and quite possibly deflates relative stats.

Other than obvious cases that have clear historic reasons for particularly bad LQ, this may very well be more trouble than its worth.

I suspect my research is closing in on illustrating that most leagues since 1901 have been pretty darned similar. This is so because baseball expanded only when there was a means for it to expand successfully. When they opened the negro league market they realized they could add teams...and when they started tapping Latino players...they expanded again to match it. For a brief period from the 70s into the 80s, not only was their competitive balance because free agency allowed for teams to break up the dynasties while the salaries were still reasonable, but there was unparalleled depth due to the baby boom generation and the latin wave overwhelming the game. For the most part...I am now convinced that aside from the 70's/80's bulge and brief flashes of LQ drops in the expansion years, in inferior leagues and prior to 1900, baseball has been very stable despite a constnatly changing player landscape.

[SABR Matt]

A little update...

I'm working on doing the same two things for the pitchers now (finding the average career length for a pitcher in a league and then finding the league's tendency for performance anomalies)...I should have that done within the hour...

Right now I'm thinking that synthesis of these variables will be more complicated than simply rating each one and taking some kind of average for the whole league. I think that the skew term tells us how hard it is to separate from the league, and the other two terms tell us something about the baseline for the league's actual ability. I'll take a look at the separate ratings and bring some of that info here for discussion.

[SABR Matt]

Problem...

Pitching usage has dramatically changed with time.

The advent of specialized relief pitchers and the death of the one man...two man...three man...and four man rotation has led to a general drop in average length of a pitcher's career in DIO...kinda blows up my method.

The win fluxes still work and look pretty similar to the offensive win fluxes although smaller in magnitude because hitters get more wins than pitchers do so things are slightly magnified. But I'm not entirely sure what to do about the changes in pitcher usage.

[jalbright]

I don't know if this will help, since you're rating everybody, but I'll throw it out just in case it does. I found my rankings of pitchers were a mess using win shares over this very issue. I've decided to keep career win shares, but for the peaks, I'm going to use linear weights as ditching the IP aspect of win shares, and then the modern guys who pitched a lot less innings can match up better on peak. In win shares, there's approximately one pitching win share per 18 innings (one team wins and gets 3 win shares, and pitching winds up being about 1/3 of that), so if in your measures you could adjust by pulling out at least some of the "average" performance based on eating innings, it might help.

Jim Albright

[SABR Matt]

Well I've already found ways around that problem you're talking about but the problem I'm having is that the very thing I'm trying to measure *IS* innings.

Just as I used PA to measure "turnover" in the league's player pool as a sign of a weak league...I was trying to use innings (outs) to measure turnover in the pitching pool...but I'm realizing that won't work for pitchers because even when they sucked, there were very few people who pitched in the 19th century so teams were kind of stuck with them.

Incidentally...the problem you're talking about...the win share problem for early pitchers...is one of the major reasons I don't care for the Win Shares system when it comes to defensive analysis. While those pitchers did cover more innings back then...they were more or less live action batting practice pitchers. In the early game they were required to throw the ball where the batter wanted it...and even when that went by the wayside, their job was to throw it over the plate and let the fielders handle all of the contacted balls. Until Cy Young showed up, no one even considered the possibility of pitching in the aggressive dominant way he did. Since PCA is based on DIPS analysis, the pitchers who would be getting way too many wins are getting a lower win rate because all they did was put it in play which is not a pitching skill.

In any event...I'm thinking that measuring the minor-leagueishness of the pitchers in the pre-1900 period is hand-in-hand with measuring the skew of the run scoring distribution so I'm dropping the Defense-Independent Outs componant of the era adjustment prior to 1901 and, going further...I'm rating 20th century (and 21st century) pitching minor-leagueishness in seperate elements...one for starters and one for relievers, which will only begin in 1950.

Here's an interesting thing I'd like to add to my conglomerate era rating method...someone on this site once mentioned that they rated the era based on the ratio of earned runs to unearned runs...which I think is a very good way to eatimate the minor-leagueishness of the FIELDERS...which should be included (bad fielders make mistakes...often in the form of allowing hits...MINOR LEAGUE fielders will tend to make many more visible mistakes...which generally get scored as errors).

[jalbright]

If all you're trying to do is measure how many innings guys had left in their careers, it's interesting to note that despite the usage patterns, career IP isn't that much in flux--probably in large part to the tendency of baseball to push pitchers to the limits of their endurance no matter the circumstances. In the old days, they just seemed to get there in less years.

Jim Albright

[SABR Matt]

That's what I thought too...that's why I thought measuring usage of pitchers would be possible...but back in 1880, the average pitchers in the league lasted 7000 innings and in 1980 it was more like 4500...largely due to relievers I think.

[jalbright]

Could the problem in the 1880's be at least partly due to the fact so many fewer pitchers were needed, thereby helping teams pick more from the cream of the crop? Now, teams need five starters, a closer, a set up man plus three or four more pitchers. Back then, if a team used four with any regularity, it was using a lot of pitchers.

Jim Albright

[SABR Matt]

Yeah...that doesn't really tell me something useful about the 1880s in terms of the quality of the league though.

I experimented with different ways of combining the four componants I've come up with so far for an era measurement:

Position Player Lifespan
League Win-Creation Anomalies
Earned-Run percentage (of total runs)
Skew of the RS distribution

last night, but didn't like what I was getting by rating each element and then playing different combinations...the results didn't frequently make as much sense as I'd like.

The RS skew is...kind of...addressed by the FSIA. That's because teams get less credit for one-sided blowouts when they're rated by the FSIA. Probabilistically, the likelihood that if the game were replayed under the same conditions the score would again be a blowout is small, so when you go game by game and analyze the real strengths of each team, in blowouts the gap tends to shrink a lot (if you win by 15, you often get credit only for being like 8 or 9 runs better than your opponant in that game)...the FSIA also (more of less) adjusts for the imbalance in teams. This is because the strength of schedule of the opponents you face is fully accounted for, so if you're on a good team, the FSIA sees you as facing an inferior league (often inferior by as much as half a run per game).

The win-creation anomaly of a league can be adjusted for. I could simply tweak all of the normalized wins in a league and tweak them down or up for all players by the same rate to account for those anomalies. That adjustment would be very small.

That leaves me measuring the minor-leaguishness of a league as my only remaining era adjustment.

Player turnover is one factor, but it's not the only factor in determining minor leaguishness.

One possibility would be to grab all of the players who lasted longer than the average career length...and all players who lasted less than the average career length...for each league and take a look at the win-scoring characteristics of those players to see if there's a pattern.

If Ubiquitus is right and there's more of a tendency for bad players to get long careers anyway because the league is so thin that teams are forced to do this...we might expect the "longer-than-average" group to do worse than normal relative to the "shorter-than-average" group. Then again...the other players who had long careers have more sucky players to beat up on for longer periods of time...so perhaps we might find something by looking at the spread of performances between the best and worst players in the long-career group? Of course PCA is already normalized for the entire player pool, but this is a particular subset so we might see something...

Will have to play around some more when I have more time.

[jalbright]

That leaves me measuring the minor-leaguishness of a league as my only remaining era adjustment.

Player turnover is one factor, but it's not the only factor in determining minor leaguishness.

One possibility would be to grab all of the players who lasted longer than the average career length...and all players who lasted less than the average career length...for each league and take a look at the win-scoring characteristics of those players to see if there's a pattern.

If Ubiquitus is right and there's more of a tendency for bad players to get long careers anyway because the league is so thin that teams are forced to do this...we might expect the "longer-than-average" group to do worse than normal relative to the "shorter-than-average" group. Then again...the other players who had long careers have more sucky players to beat up on for longer periods of time...so perhaps we might find something by looking at the spread of performances between the best and worst players in the long-career group? Of course PCA is already normalized for the entire player pool, but this is a particular subset so we might see something...

Will have to play around some more when I have more time.

Minor leaguers would definitely be in the bottom half, third or even quarter in terms of production (offense and defense combined), so if the bottom quarter (maybe even less) are having longer careers per Ubi's argument, we should expect that group to have longer careers if (or when) Ubi's argument is correct.

Jim Albright

[SABR Matt]

Hmm...

I'm thinking that a thin league will be forced to play a few types of players more...

1) Rookies (players with little or no prior experience)
2) Oldies (players who are no longer productive but who are playing on reputation)
3) Specialists (players who are good at one thing and terrible at the others...good fielders who can't hit, good hitters who can't field, good baserunners...)
4) Players who have already been proven to suck in the big leagues who are still inexplicably getting PT (Bad teams like the 2005 Mariners are forced to run a starting rotation of Ryan Franklin (proven sucky pitcher), Gil Meche (pock-marked youngster with limited success), Felix Hernandez (brilliant prospect with no experience), Joel Pineiro (proven sucky pitcher), Aaron Sele (crusty old guy), Jamie Moyer (REEEALLLY crusty old guy), and stopgaps from AAA).

One can look for leagues that start more rookies by looking for the percentage of the weight of the league's PT given to players with less than a certain threshold of previous playing time.

One can look for leagues that start too many old guys by finding the percentage of the league's PT by weight given to players who are older than a certain age.

One can look for leagues that start proven suckfests by looking at the percentage of the league's PT (again by weight of PT) given to players who prior to the season in question have scored at or below a certain win scoring rate prior to the season in question.

Specialists would be harder to isolate...but I'm thinking that for position players at least, you could look for players who earn almost all of their career value in one componant or the other (fielding or hitting)...or we could do something a little more direct and look at earned-run percentage to scan for bad fielding (or something like Bill James' E/DP...which looks for the ratio of bad fielding to known good fielding) to pick up that element and let the other steps catch the hitting side.

[jalbright]

The first three elements make sense to me.

The specialists: the only real pitching specialist I can think of that we'd want to get is the lefty one out guy, and nowadays most teams want to carry one of those. As for the hitters, the DH screws up using the percentage on fielding versus hitting--at least some DHs could field if they had to, but it's better for their health and thus for the team if they don't. So I think I'd try to pick up the good field no hit types with a good fielding rating but lousy production with the bat, and the good hit-no field guys with lousy fielding ratings (if they can't field, they'd better hit to stay in the game). I think you could also eliminate corner OF, 1B and DHs from consideration in the good field, no hit category.

Jim Albright

[Ubiquitous]

My argument isn't that the bottom quarter is made up of bad long career guys but that there were bad long careers there. The bottom will still have the same cast-off type system we see today. Players playing 1 or 2 or 3 seasons. Not getting a lot of playing time and so forth. If you didn't have a glove and you couldn't hit you were not going to last no matter what era.

But to me these players who could field somewhat but had no bat are minor leaguers in that good fielders are easily replaceable. These guys were not Ozzie Smith or Bill Mazeroski with the glove. They were decent fielders that lasted awhile because finding a player who could field and hit was next to impossible for some of these teams that didn't have the resources to find or sign these players. Which is why to me you see a lot of blackhole bats up the middle for so long.

My view is that these teams were more then willing to employ minor league players on their rosters and if they got along with management played the game the way the manager wanted and had a decent glove he could last a long time. But the bottomline for me isn't that they had longer careers or shorter careers or whatever length of career but that they were there and there were a lot of them.

Now it is certainly possible that due to long term contracts and the growing disparity of revenue streams we are starting to see more players who shouldn't be in the majors. Never said it wasn't possible, but that A) it is harder for these players to have as great an impact on a league B) teams show a greater reluctance or ability to maintain minor league futility for as long as past teams did and C) even teams that might be minor league quality overall generally don't have as many minor league players that teams from the past did. Meaning they have a few players who are either clearly major leaguers or who definitely have a future in the game, while teams like the Braves, Browns, or A's had almost nobody on their roster who was a major leaguer or was going to be.

[csh19792001]

If all you're trying to do is measure how many innings guys had left in their careers, it's interesting to note that despite the usage patterns, career IP isn't that much in flux--probably in large part to the tendency of baseball to push pitchers to the limits of their endurance no matter the circumstances. In the old days, they just seemed to get there in less years.

Jim Albright

Jim,

In a discussion of Cy Young's move to the AL after signing the contract in March of 1901, Browning discusses this exact topic, framed in historical context.

"In the end Frank Robison (owner of the Cardinals, among other teams) may have come very close to meeting Boston's offer- in dollar terms Robison may have actually topped it. But Robison would not go beyond a one-year committment (Boston offered a three year deal), and Stanley Robison explained the club's reasoning- Cy Young was in decline and would probably not last more than one more year in the National League.

It is easy for us to laugh at Frank Robison's poor judgement. But in fact he had much actuarial experience on his side. Cy Young celebrated his thirty-fourth birthday about the time he signed the new contract. How well had pitchers past their thirty-fourth birthday fared over the past quarter century? Charley Radbourn had done best- 58 wins and 36 losses from the age of 34 on. A few others had continued to be effective enough to hang on in the majors- Pud Galvin with 36 wins, Tim Keefe with 34, Tony Mullane with 25, and Bobby Mathews with 16. But that's about it. Charlie Buffinton, John Clarkson, Silver King, Jim McCormick, Mickey Welch, Jack Stivetts, and Will White had hung up their gloves by the time they were 34. Thus, Robison had history on his side in declining to extend Young a lucrative, mult-year contract."

-Cy Young, A Baseball Life (Browning, p. 92)

[csh19792001]

PS- This historical fact about pitcher usage is going to distort any metric that claims to (directly) infer league quality from career length- pitchers were used sometimes twice as much in the 19th century as they are today- and it was physically impossible to last till or past 40 under those conditions, with no training or modern medicine. This usage patterns resulting in truncated careers is irrespective of the actual quality of the league.

It's very just to impute numbers in, get numbers out, and draw hard, absolute conclusions from them. But it neglects all of the historical context that allowed those numbers to occur, thereby yielding completely erroneous and misguided conclusions.

[SABR Matt]

Agreed csh...which is why I immediately threw out the data I got back from the usage study. The pitcher usage pattern in 1880 was so utterly and radically different than it is today that it was impossible to draw hard conclusions.

I am also essentially throwing out my batter PA study because there are problems with the concept that under normal, non-war-stressed conditions, the league will have a shorter life expentency if it is weak. Players on weak leagues find it easier to continue to get jobs with limited skill and also find it easier to make themselves appear to be subsisting statistically because they're not facing elite competition. The lifespan argument can catch massive fluctuations in the talent pool but it won't catch the likely continuous minor changes in overall ability of the league.

I'll go ahead and attempt to research "old crusties", rooks, and sucky hangers on and see if I can find some patterns there...then we'll look at the no-hit good glove types and the good bat no field types (with DH, 1B, and LF/RF removed from consideration to avoid confusing the issue).

[SABR Matt]

I've been thinking a little bit about the usage patterns that signal a weak league.

I listed four elements that show a league is weak...but I think that can be simplified to two.

On the one side you've got players with little or no experience. Check. That is its' own variable and can easily be diagnosed independently for batters, pitchers and fielders.

Now this idea that leagues use specialists, old veterans, and ineffective players in general more often if they're bad leagues I think can be captured in one variable. It's not bad for the league to start a 40 year old if his age 37-39 seasons were all productive. It's not bad to use specialists a lot of they perform their special functions correctly (as someone above pointed out...using a LOOGY is fine as long as your LOOGY pitchers get batters out and help teams win). The bottom line is...bad leagues will be forced to use players who have a recent history of ineffective play more frequently than good leagues. The thing that made the 70s and even 60s unique in the history of the game was that rather than relying on guys you knew would suck, a lot of teams went out and bought Negro League players (and later on...latino players)...today, teams are digging through Asian prospects to fill those last roster spots that would get filled with sucky underperformers in a weaker league.

I think red flag #1 for a bad league is...percent of PT by weight of players whose last X PA or IP or games in the field were statistically weak.

If you combine the percentages of PT by weight of players with little or no experience with the percent of PT by weight for players who have not recently been effective...you'll have an idea of how much leagues are being forced to rely on players that are "high risk"...either from a lack of experience or from a lack of skill.

[csh19792001]

I've been thinking a little bit about the usage patterns that signal a weak league.

I listed four elements that show a league is weak...but I think that can be simplified to two.

On the one side you've got players with little or no experience. Check. That is its' own variable and can easily be diagnosed independently for batters, pitchers and fielders.

Now this idea that leagues use specialists, old veterans, and ineffective players in general more often if they're bad leagues I think can be captured in one variable. It's not bad for the league to start a 40 year old if his age 37-39 seasons were all productive. It's not bad to use specialists a lot of they perform their special functions correctly (as someone above pointed out...using a LOOGY is fine as long as your LOOGY pitchers get batters out and help teams win). The bottom line is...bad leagues will be forced to use players who have a recent history of ineffective play more frequently than good leagues. The thing that made the 70s and even 60s unique in the history of the game was that rather than relying on guys you knew would suck, a lot of teams went out and bought Negro League players (and later on...latino players)...today, teams are digging through Asian prospects to fill those last roster spots that would get filled with sucky underperformers in a weaker league.

I think red flag #1 for a bad league is...percent of PT by weight of players whose last X PA or IP or games in the field were statistically weak.

If you combine the percentages of PT by weight of players with little or no experience with the percent of PT by weight for players who have not recently been effective...you'll have an idea of how much leagues are being forced to rely on players that are "high risk"...either from a lack of experience or from a lack of skill.

One caveat with looking at the number of old players past 40 (which I'm not sure you're incorporating, but it has come up) is that with the advent of much better training/personal care (particularly in the past 10-15 years), there are simply going to be far more older players in the league. You probably already consider this, but just in case.

In other words, the average player age hasn't gone up steadily simply due to college and more player development in the minors; it's added in on the tail end as well. The average age of the major league player in the 19th century was 23-25 years old, today it's 28-29. Is that an indica of quality in and of itself?

I think if you only look at age in relation to performance, then it shouldn't confound things.

Also, what's your stance on the dramatic increase in relief pitching and specialization over the past 10-15 years as indicative of a dropoff in league strength (or, perhaps, difficulty for hitters). It's probably greatly oversimplifying things, but just eyeballing the splits for starting pitching vs. relief pitching, it looks like relief pitching does significantly cut into hitting in relation to starting pitching.

Wouldn't that mean that the more relief pitching is used in a league, the more difficult it will make things (in general) for hitters?

You'd have to first look at the trends in starter vs. relief ERA and correlate it with the percentage of innings pitched per year by relievers. That might yield some very telling information.

[SABR Matt]

Yeah Chris...that's one of the reasons I changed my mind and switched away from looking at chronological age and decided to look simply at players who are somehow still playing despite diminishing returns. I'll look specifically at the percentage of players who have a significant period of ineffectiveness under their belt and are somehow still getting PT.

[redbuck]

Matt, These are great.

Am I correct that these are usable scores- that they can be used against each other?

For example, can I apply these to ratings so that a 1907 stat is only 651/968 as "good" as a AL 1984 stat?

[SABR Matt]

Matt, These are great.

Am I correct that these are usable scores- that they can be used against each other?

For example, can I apply these to ratings so that a 1907 stat is only 651/968 as "good" as a AL 1984 stat?

That's the theory...the currently available era difficulty rating is not IMHO an accurate measure all by itself, but when I get a working EDR, that's how I intend it to be used, yes.

If 1876 is .750 and 1976 is 1.030 then to put the players on the same difficulty field, you would just multiply each player's wins created or OPS+ by the era difficulty.

If a player earned 10 wins in 1976 he'd be worth 10.3 wins in a neutral context...whereas he'd be worth 7.5 wins in a neutral context if he earned 10 wins in 1876

[csh19792001]

Yeah Chris...that's one of the reasons I changed my mind and switched away from looking at chronological age and decided to look simply at players who are somehow still playing despite diminishing returns. I'll look specifically at the percentage of players who have a significant period of ineffectiveness under their belt and are somehow still getting PT.

What about your thoughts the advent of relief pitching/specialists, and how it pertains to league strength?

What about this endemic change- in the long run, do you believe it makes it more difficult for hitters of the past 15 years juxtaposed with, say, guys from the 50's-70's?

If we were to find that relief pitching over the past 50 years has been significantly more effective than starting pitching, would that automatically lead us to the conclusion that the more relief pitching is incorporated, the more difficult it has become for hitters to perform to the same standards as they did in eras bygone?

And since it's been pretty much a linear move towards less innings/more pitchers since 1871 (with the exception of the second deadball era), what historical/statistical implications might this carry?

[SABR Matt]

On the question of relievers, yes, I believe the specialization of relief pitching has improved the pitching game above what it might otherwise have been...however, since the depth of pitching has recently (most likely) declined in the 90s, there may be some counterbalancing going on. I do believe there is evidence that allowing SPs to throw fewer innings has improved their pitching...and using talented specialists to finish games is making it harder for hitters (more pitchers to scout and adjust to, many of whom have better stuff than a starting pitcher would by the time he got to the 8th inning...plus...your arms are more fresh and less prone to injury).

As for the more or less linear decline on the overuse of starting pitchers...I would say that may be acting to improve the quality of defenses and thus the overall difficulty of the league slightly, except that there is some sabermetric evidence that the 5th starter is a bad idea (because the talent pool isn't deep enough so you're giving a lot of innings to a mediocre minor leagueish pitcher most of the time)...so...I guess that data would have to be studied a little more in depth.

[jalbright]

Anything happening with this research in the past week or so, Matt?

Jim Albright

[SABR Matt]

My computer has been having issues, so I've been hesitant to do anything resembling serious research on it in case I suddenly have to replace the hard drive.

I need to remind myself what I was planning on doing and get back to it once exams are over (this Wednesday evening is my last exam...I get home Friday night.

[redbuck]

The ratings from 1943-45 are here and actually pretty high. Is there any way these can account for the loss of talent during WWII?

And I had heard that 1961 was a pretty weak year in baseball. I don't know if I'm right or not, but it shows up pretty high on the list.

[SABR Matt]

Remember...I've already said in this thread that the skew-based ratings published in this thread do not IMHO represent a full accounting of the strength of the league. I agree with you that the WWII depression is not enough in the skew ratings and that it's likely 1961 and 1962 (and 1969 and 1977) were weaker than the skew ratings showed.

[Windy City Fan]

Any further progress on this front? I find your efforts here to be really valuable.

[SABR Matt]

I am now of the opinion that an era adjustment should be about fitting players to the same quasi-normal distribution, and adjusting linearly only when there is significant evidence of minor-leagueishness. As I've said in a couple of other places, I don't think the solution to the era question is to come up with a linear scale for each season (without making other changes) because the goal is to level the playing field, and the only way the playing field can be leveled is by fitting players to the same distribution of performance much like Schell does.

Step 1) Find the distribution of PCA win scoring rates or WS/PA or WARP3/PA...whichever stat you prefer (I'm using PCA because I have the data) in each league/year.

Step 2) Use a power transformation to make that distribution less skewed and more normal. I'll need to experiment with different power rules (a power transformation sounds scary but it's just taking your data and raising it to some power...if your distribution is highly skewed right, you need a power of less than one...that way large values are reduced much more than small values and you get a more bell-curve like shape) to find which power comes closest to normalizing the data in different time periods.

Step 3) Make a further correction to the power-transformed data to force the standard deviations for each league/year to be equal.

Step 4) Trnasform all data back to proper scale by undoing the power correction...except this time, use only one standard (the standard used to correct data from the 1970s and 1980s will probably be used because IMHO the 70s and 80s were the most balanced leagues)...this way all data will be shaped the same way.

Step 5) Test for leagues which do not belong on the same scale (leagues which show blatant evidence of being more minor-leagueish) and make a linear adjustment to those seasons. I still think the way define minor leagueishness is through a search for leagues which (a) give players with a recent history of poor performance more playing time and (b) give more playing time to untested rookies.

The drawback of this approach is that I can't simply publish a table that shows what era adjustments need to be made and start advocating that chart's use in debate because now I'm working with whole distributions instead of just trying to put a number to a league's strength, so in order to use any era adjustment I make...you would need to be working with the same data set as I am and applying all of the same filters. And of course, I suffer from having to convince people that the data set I would use is a valid starting data set (that PCA is good enough to use as the starting point), and that all of my statistical machinations to get players on a levle playing field are justified and succeed in accounting for changes in era.

I'm working on setting up a file in Excel where I can test raw win-created data and see if I can get a good method for picking power rules now.

[jalbright]

I think I hear what you're saying, but a major drawback with this is that until we get to "kick the tires" of the finalized PCA system with a real chance to dig through it, we're not getting anywhere. Perhaps at least you could tell us a) which leagues are so bad as to require a linear adjustment, and b) some ranges of the size of adjustment for certain periods of time.

Jim Albright

[SABR Matt]

Yep...I know...that's why I called it a drawback in the last paragraphs of my last post.

PCA is not presently in a form I'm 100% happy with as is. My problem right now is...there is so much data to integrate and I am so ill equipped to handle such a complex database (integrating data from the baseball-databank, the retrosheet gamelogs and PBP event files, the KJOK Parks database and the FSIA Matrix solution set, along with possibly the HR database...is a monumental job...but all of that data is needed to get what I want to get done...done. I've spend a lot of time nad money lately attempting to educate myself in database management, but I can't even wrap my head around how to get all of that data to coexist without blowing up my computer simply by opening the master database let alone how to carry out complex math on the DB (I'm not a computer programmer).

[jalbright]

Databases are wonderful animals when they're tame enough for you to work with to get what you want, but when they're beyond you, the results are often as gruesome as I imagine you'd see if I tried to take over the lion tamer act at the circus (solo--or would it just be: so long:waving ?):ughh . Been there, done that.

I may have a cheat sheet answer for you, though. Let's face it, most folks are interested in the stars. What if you published the results by league for the 20th or 25th best guy or at the 95th or 97th percentile, with a disclaimer that the further you get away from that quality, the less applicable the adjustment would be? It might not be mathematically precise, but I'd think it would be close enough for most people in the range of players they're really interested in.

Jim Albright

[SABR Matt]

Well that doesn't really simplify the process. I still have to get 6 different data sources to talk to each other to do any calculation of ratings.

I was however considering limiting my study of scoring rates to "regular players" like Schell does in his book.

[Windy City Fan]

Unfortunately, you're talking waaaaay over my head now. Are you saying it's impossible to say, "the NL in 1905 has a X modification factor". And if so why, and what kind of modifiers/numbers are you hoping to produce for us to use? How would the system your contemplating work for adjusting for leage depth?

(Keep in mind when you reply that I'm not well versed in advanced SABR or statistical methods and terminology)

[SABR Matt]

Sorry Windy City...:) I didn't mean to confuse ya.

I'm saying I am not convinced that the correct way to adjust for league depth is with a linear adjustment, except in obvious cases where there was high player turnover (the dawn of baseball, world wars, new leagues, and maaaybe the biggest expansions.

I think adjusting for era is going to come down to that kind of linear adjusting for extreme cases combined with a fuller analysis of the distribution of player performances. Unfortunately, that kind of analysis doens't result in nice numbers I can publish for you guys to use as rough multipliers...the way I may adapt it for more general use is by figuring out what to mutiply certain skill levels by to get a close approximation to my era adjustment.

Example:

Let's say that correcting for the extremity of player seaosns in a weak league results in a 30 WS player becoming a 22 WS player in 1905. I can figure out what the conversions are for key levels and publish THOSE for each season. It won't be one linear number...I won't call 1905 a .73 league just because the 30 WS player became a 22 WS player, because chances are...the 20 WS player didn't get reduced all the way to 14 WS...the closer you get to the mean...the less the adjustment.

[jalbright]

Matt,

Can you square that approach with the idea I understand is favored in statistics of seeding from the top? I don't see how your approach fits with that idea, but I may just be missing something.

Jim Albright

[SABR Matt]

seeding from the top? Be more specific if you could as I'm not sure how that applies to this conversation.

[jalbright]

I wish you had the documentation I have, a booklet that goes with an APBA Eastern Colored League set. They did the usual compilation of Negro League stats, then deduced what they could, and finally filled out the data with reasonable estimates. Their goal was to make the set playable against some sets from the same time period. I think Oscar Charleston's unadjusted stats gave him a .400+ avg, which was about 6% higher than the highest average of a full time player in the (IIRC) 1924 set they used as their comparison. So, Oscar was moved down to about that level. The best HR hitter would be compared to Ruth, etc. Of course, then everybody moved down, which meant that the mean of the Negro Leaguers was below the mean of the major leaguers. As I understand what they said, unless there's good reason to reject the notion, the best of one time or place are comparable to the best of another time and place (obviously, when one sprinter runs the 100 meter dash in a second less, that's good reason to differ--but baseball isn't such a single skill event). If you want jpgs of the pages and you can give me an email via PM, I can scan and send it to you.

Jim Albright