What value is there to wins and losses?

[George H Ruth]

What value is there to wins and losses? I can't think of anything solid.

[mikefast]

I never found much value in losses. Yeah, sure, no pain, no gain, yadda, yadda, yadda...

Wins, on the other hand, are usually a lot of fun. I recommend them.

[thefeckcampaign]

I find the win has turned into an unappreciated statistic and maybe rightfully so. I honestly do not know enough about most of the new stats to really consider what is valuable anymore. I will say though with the choices of Cy Young Award winners today, the times have changed.

I think perhaps that pitchers are being taken out too early today and ultimately devaluing the WIN. Sure there were always some pitchers that road the coattails of being on great teams (Don Sutton), where some were not so lucky (Bert Blyleven) so the win definitely does not tell the whole story.....

But looking at Steve Carlton's 1972 season with the Phillies, he pitched 30 complete games and won 27 of them on a team that only won 59 games that whole year. To me, that says a whole lot regardless of what the other numbers might say though they are impressive as well.

I think if Zack Greinke would have finished as many games as he was pitching well (18 with 1 or less runs) you would have seen a lot more wins than 15 and his team would have had a better record as well. That is if he continued to throw a 1 run or less game through 9 innings.

Now it seems he averaged about 6.9 innings a start, so someone was blowing the game in the late innings or the Royals were flat out getting shut out. Now the main question is why didn't he finish those 1 run or less games in which he was only 4-2? I honestly do not feel it's anyone's fault but his own being he plays in the American League.

I just wonder that if he completed as many games Carlton, would he have had more wins or would he have had a worse ERA/season in general?

If more great pitchers like Greinke finished their own games when they were pitching well, would the Win have been as devalued as it seemed to have been by most modern statisticians or would it show these pitchers are not really as good as we think they are because they end up blowing their own games in the late innings? (Hence their other stats would go up as well, ERA, hits etc.)

Are pitchers today taken out before they blow the lead where in the old days they were only replaced when the going got rough?

[STLCards2]

What value is there to wins and losses? I can't think of anything solid.

I hate winning! Losing is way better!

[Second Base Coach]

This 'win bashing" trend of late is getting out of hand. While it leans towards being overrated, it certainly isn't worthless. It was an excellent statistic when almost every starting pitcher finished every one of his starts. It was still terrific when we had four man rotations and many pitchers finished well north of 250 innings pitched.

Now.... more and more pitchers are earning wins by pitching closer to the five inning minimum requirement than the nine innings on the day's schedule.

While it is a bigger feat to go eight innings and keep your team ahead, pitchers still have pitch WELL enough, and LONG enough to meet the requirements for a win.

I think what has happened is this... we have more stats to go with those wins. The weight on the statistic isn't what it used to because we have so many more ingredients to put into our Cy Young soup.

I would not discount wins right out of existence because there is still GREAT value in a pitcher who can pitch five-plus innings and leave with the lead.

While we may long for the nine inning ACE, the eight-inning workhorse, or even the seven-inning "innings eater" we take what we can get because fans and stat heads cannot effect how the game has changed. We aren't in charge of the evolution of the game.

What bothers so many is the arguments which are made that someone who won 14 or 15 games had a better season than someone who came in with 20. That argument used to be hard to make, when most guys left so few unplayed innings left on the scoreboard when they departed the mound for the last time. Now we count outs and look at rate stats more than ever. We now look at individual results on a per batter basis, rather than on a per game basis. So be it. Everyone is playing under the same conditions so this should not be that difficult of a process.

I still believe total wins to be a telling statistic. That does not make me nostalgic does it? You just have to place won-loss records along side everything else. We now have more of "everything else" than ever before and that is a good thing.

We just cannot do anything about how so many eight inning wins have changed into six inning wins, or worse yet, no decisions. Oh well....

[leewileyfan]

The way I look at pitchers getting credit for Wins is through the lens of baseball and sports history here in the USA generally.

From one-on-one gunslinger contests; to boxing matches between the likes of Jeffries & Johnson; the early days of tennis singles; and man-on-man match play on the links, crowds are always drawn to the concept of a winner and a loser. Even famed 2 horse challenges have been historic crowd draws.

Back when pitchers were expected to pile up CG, the morning line of starting pitchers was as much promotional as it was statistical. [Our guy vs. your guy].

It certainly does no harm, even though modern starters are basically 5-6 innings guys, with a seven inning per game guy being considered a workhorse. Hitters have a ton of stats to puff up their reputations; so why not let the media continue to toss up wins on the marquee for pitchers?

For avid fans, that morning line for starters is still a big drawing card.

[Second Base Coach]

Well said. If I have tickets to a game, I always look forward to seeing who are the starting pitchers. About five days out I try to guess, then about two days out I can generally find out.

And who among us has never turned on the radio while approaching the stadium to find out who is hurt and who might be taking his place that day. The joy of your day often depended on the quality of your team's spot starter. Imagine Don Sutton's start being pushed back a day and you got Charlie Hough instead. :noidea

I can remember watching Lonnie Smith one game, on a day he gave Greg Luzinski a day off. He really DID look like he way playing left field on roller skates. Garry Maddox would talk to him nearly every half inning when they ran off the field and into the dugout.

I wonder if Lonnie was listening...

[mikefast]

But looking at Steve Carlton's 1972 season with the Phillies, he pitched 30 complete games and won 27 of them on a team that only won 59 games that whole year. To me, that says a whole lot regardless of what the other numbers might say though they are impressive as well.

I think if Zack Greinke would have finished as many games as he was pitching well (18 with 1 or less runs) you would have seen a lot more wins than 15 and his team would have had a better record as well. That is if he continued to throw a 1 run or less game through 9 innings.

Now it seems he averaged about 6.9 innings a start, so someone was blowing the game in the late innings or the Royals were flat out getting shut out. Now the main question is why didn't he finish those 1 run or less games in which he was only 4-2? I honestly do not feel it's anyone's fault but his own being he plays in the American League.

I just wonder that if he completed as many games Carlton, would he have had more wins or would he have had a worse ERA/season in general?
Greinke's record when the Royals scored zero runs was 0-3 in 4 starts.
Greinke's record when the Royals scored one run was 0-1 in 2 starts.
Greinke's record when the Royals scored two runs was 3-1 in 6 starts.
Greinke's record when the Royals scored three runs was 3-1 in 7 starts.
Greinke's record when the Royals scored 4+ runs was 10-2 in 14 starts.

Carlton's record when the Phillies scored zero runs was 0-2 in 3 starts.
Carlton's record when the Phillies scored one run was 2-3 in 5 starts.
Carlton's record when the Phillies scored two runs was 7-2 in 9 starts.
Carlton's record when the Phillies scored three runs was 4-2 in 6 starts.
Carlton's record when the Phillies scored 4+ runs was 14-1 in 18 starts.

Carlton started eight more games than Greinke, and nine more than Greinke in which his team scored a run. So that's part of it. But if Greinke had won games at the same rate Carlton did given each level of offensive support, he'd have won 21 games instead of 16. That's one way to look at it.

Another way to look at it is that the bullpen blew three games where Greinke left with the lead.

[thefeckcampaign]

Greinke's record when the Royals scored zero runs was 0-3 in 4 starts.
Greinke's record when the Royals scored one run was 0-1 in 2 starts.
Greinke's record when the Royals scored two runs was 3-1 in 6 starts.
Greinke's record when the Royals scored three runs was 3-1 in 7 starts.
Greinke's record when the Royals scored 4+ runs was 10-2 in 14 starts.

Carlton's record when the Phillies scored zero runs was 0-2 in 3 starts.
Carlton's record when the Phillies scored one run was 2-3 in 5 starts.
Carlton's record when the Phillies scored two runs was 7-2 in 9 starts.
Carlton's record when the Phillies scored three runs was 4-2 in 6 starts.
Carlton's record when the Phillies scored 4+ runs was 14-1 in 18 starts.

Carlton started eight more games than Greinke, and nine more than Greinke in which his team scored a run. So that's part of it. But if Greinke had won games at the same rate Carlton did given each level of offensive support, he'd have won 21 games instead of 16. That's one way to look at it.

Another way to look at it is that the bullpen blew three games where Greinke left with the lead.
These are great stats and I appreciate you posting them. I am amazed how similar they are but I do not really follow the bold. Are you saying if he had as many starts as Carlton at the rate he was going he would have ended up with 21 wins?

It still makes me wonder why he was taken out of so many games if he was pitching so well. Why wouldn't a team with such a questionable bullpen just have him finish? He only had 6 complete games all year which compared to Carlton's 30 or even Don Sutton averaging 14 or so seems weak.

What kind of season would he have had IF he could have finished his own game and gotten his own win.

It seems to me from this discussion the WIN shows you are a work horse. You finish your own games and you tend to start more often. You are basically caring your team more. No?

[Second Base Coach]

Or at the very least, you are sticking around long enough to get the decision (win or loss).

[leewileyfan]

I think what has happened is this... we have more stats to go with those wins. The weight on the statistic isn't what it used to because we have so many more ingredients to put into our Cy Young soup. ....

I imagine that stats expand as varieties of "specialization" arise, which is probably a necessary corollary of game evolution. However, the expanded stats are a reaction, not a cause.


While we may long for the nine inning ACE, the eight-inning workhorse, or even the seven-inning "innings eater" we take what we can get because fans and stat heads cannot effect how the game has changed. We aren't in charge of the evolution of the game.

While "fans and statheads" cannot effect changes in the game, adult participation, especially those fulfilling failed childhood dreams, now transferred to little kids for realization certainly can and have changed the game - and not always for the better.

These observations are my own; and I won't be shocked by any negative responses they may invite; but I am convinced they are valid:

1. Little League, for all the good it may have done for youthful participation and inflation of young egos given the opportunity to play in "big time" atmospheres, had left a heavy adult footprint on these young competitions:

a. Once LL really began to catch on, expanding through the Western Hemisphere and onto the global stage, it got media attention; and the focus shifted from kids to adults and the very adult concept of winning, on a plane at least equal to the stage on which one performed.

b. In this climate, a Petri dish for ringers being brought in to represent "teams" that fared well in neighborhood or lesser regional competitions. Just recall the many years of dominance of teams with pitchers at or over 6' tall and flaming baseballs 40', looking more like 16-18 year olds than LL prospects. Some awesome hitters, too, seems closer to five o'clock shadow than to kindergarten.

2. As questionable practices were addressed, aluminum bats came on the scene, giving kids, their coaches and their parents visions of bat speed ans awesome power generation not genuinely possessed if a kid was tested to perform in like manner, with wood.

3. If we put the above in a time context, we get kids @ 10-13 or so around 1952-57, with a later, power-oriented wave @ 10-13 in the late 70s and early 80s.

4. I suggest the first wave accounted in large part for MLB pitcher dominance from 1965-72; and the second wave contributed greatly to the big whiff wave from the mid-'80s to the present.

5. With LL being pitcher-centered, we have coaches experimenting with young arms for competitive edge, OR, pampering young arms to such a degree that the kids learn to pitch down to low expectations, as far as stamina is concerned.

Agreed, fans and statheads have diminished impact on game evolution; but competitive adults who may squash childhood initiatives, certainly can.

[digglahhh]

Here are two of my favorite myths, misconceptions about pitchers and win totals.

Pitchers somehow control how many innings they pitch or the number of games they start. Every time a manager goes to the mound to remove his ace with two outs in the seventh, that pitcher has a look on his face like he does not want to leave the game. He tips his hat to the crowd and walks disappointedly to the dugout. We all see this, right? I mean, we aren’t lowly statheads who don’t actually watch the games and spend all our days in our parents’ basements playing WarCraft, right?

Anyway, despite the fact that we all know this people still throw out the old reliable canard of “if Greinke would pitch more innings, he’d have more wins” as if Greinke removed himself from those games. Trust me, Zack Greinke wanted to finish those games. To say that pitcher X should have pitched more innings is a straw man when evaluating the skills of pitcher X because he has very little control over how many innings he pitches in the first place.

My second beef is less of a myth and more of just an internal inconsistency in argumentation. So, if wins imply that a pitcher pitched well for his team to win, why does it matter so much that pitchers today usually pitch fewer innings? If 6 inning wins aren’t a good enough barometer of a pitcher’s success, then that seems to be a flaw with the stat, not the pitcher earning that stat. When people bash these pitchers and their six inning wins, they are bashing the win stat, but then those same people often turn around and extol the virtues of high win totals, even though they have already implicitly acknowledged the arbitrary nature of what it takes to get a win. (Also, why use high win totals as a proxy for durability or ability to go deep into games when we have very straightforward stats established just to measure those things, like, I dunno, innings pitched. And finally, the same people who bash the six-inning, three-run quality start, are the same people to assign positive value of the pitcher who goes eight and gives up five, while earning a 6-5 win.

The lore and the romance of the game is one thing, precisely measuring the performance of those who play it is something different. It’s important that people remember this.

[STLCards2]

Here are two of my favorite myths, misconceptions about pitchers and win totals.

Pitchers somehow control how many innings they pitch or the number of games they start. Every time a manager goes to the mound to remove his ace with two outs in the seventh, that pitcher has a look on his face like he does not want to leave the game. He tips his hat to the crowd and walks disappointedly to the dugout. We all see this, right? I mean, we aren’t lowly statheads who don’t actually watch the games and spend all our days in our parents’ basements playing WarCraft, right?

Anyway, despite the fact that we all know this people still throw out the old reliable canard of “if Greinke would pitch more innings, he’d have more wins” as if Greinke removed himself from those games. Trust me, Zack Greinke wanted to finish those games. To say that pitcher X should have pitched more innings is a straw man when evaluating the skills of pitcher X because he has very little control over how many innings he pitches in the first place.

My second beef is less of a myth and more of just an internal inconsistency in argumentation. So, if wins imply that a pitcher pitched well for his team to win, why does it matter so much that pitchers today usually pitch fewer innings? If 6 inning wins aren’t a good enough barometer of a pitcher’s success, then that seems to be a flaw with the stat, not the pitcher earning that stat. When people bash these pitchers and their six inning wins, they are bashing the win stat, but then those same people often turn around and extol the virtues of high win totals, even though they have already implicitly acknowledged the arbitrary nature of what it takes to get a win. (Also, why use high win totals as a proxy for durability or ability to go deep into games when we have very straightforward stats established just to measure those things, like, I dunno, innings pitched. And finally, the same people who bash the six-inning, three-run quality start, are the same people to assign positive value of the pitcher who goes eight and gives up five, while earning a 6-5 win.

The lore and the romance of the game is one thing, precisely measuring the performance of those who play it is something different. It’s important that people remember this.

Very well said! that logically fallacy drives me nuts too! Nice to see you around here - seems like it has been a while.

[Los Bravos]

Nice to see you around here - seems like it has been a while.Agreed. Always nice to read what Mr. d has to say.

[leewileyfan]

Here are two of my favorite myths, misconceptions about pitchers and win totals.

Pitchers somehow control how many innings they pitch or the number of games they start. Every time a manager goes to the mound to remove his ace with two outs in the seventh, that pitcher has a look on his face like he does not want to leave the game. He tips his hat to the crowd and walks disappointedly to the dugout. We all see this, right? I mean, we aren’t lowly statheads who don’t actually watch the games and spend all our days in our parents’ basements playing WarCraft, right?

To argue that pitchers somehow have no control over the number of innings they pitch [at least for starters, that is] begs the question in a huge way, which can be illustrated by some simple common-sense facts about pitching staffs in general:

a. Expansion his diluted the talent pool and the tendency to carry 12-16 pitchers north when spring training camps break makes the dilution excruciatingly obvious on clubs lacking a reliable standout closer.

b. A pitcher is put on a strict pitch count only when he has either illustrated early fatigue; is recovering from a strain or injury of some sort; has no sense of pacing or matured finesse in pitching in some degree to the score.

c. A pitcher, to cite your own example, who gets visited on the mound with two outs in the 7th inning, even in a deep situation, will stay in the game for that extra one-third of an inning, meaning he recorded 78% of a nine-inning game's outs. The pride you cited in your example must be earned, though. A guy with a track record of hitting the wall in the 5th or 6th will not gain that extra trust. He hasn't earned it.



Anyway, despite the fact that we all know this people still throw out the old reliable canard of “if Greinke would pitch more innings, he’d have more wins” as if Greinke removed himself from those games. Trust me, Zack Greinke wanted to finish those games. To say that pitcher X should have pitched more innings is a straw man when evaluating the skills of pitcher X because he has very little control over how many innings he pitches in the first place.

There is no canard and no straw man, so long as the clear-cut observation is taken to task honestly. MLB is a nine inning game. For a game to be official, it must go 4.5 innings. Half a scheduled game qualifying as official seems both practical and at the extreme end of generous, if we consider the vicissitudes of weather and temperature and unexpected calamity.

In making "the rotation" on a MLB staff means, quite simply that, of 12-14 pitchers, such a guy is in the elite 33% or 40% of the staff, with others in varied supporting roles, maybe 1 or 2 vying for "closer" or some like dramatic role. Thus, in a nine inning game, a "star" must go 5 innings to qualify for a win. Another "star" must be effective for three outs. We take out the other guys in the rotation [3 or 4] and we are left with 5-7 pitchers making MLB salaries to fill in 2-3 innings in a game.

Earning a win, in the entire context, does not seem too Draconian an expectation, especially when a starter has to pitch only 5 innings to qualify. Heck, in larger context, LL was founded @ 6 innings so little kids wouldn't get too physically stressed.


My second beef is less of a myth and more of just an internal inconsistency in argumentation. So, if wins imply that a pitcher pitched well for his team to win, why does it matter so much that pitchers today usually pitch fewer innings?

Logical argumentation contains no such flaw. If Clem has 34 starts and 0 complete games, he can be quite impressive IF he averages getting into the seventh inning per start.

Say his line is:

Starts.............IP/Start.........IP............CG.........QS........W ............L

34.....................6.3..........214.2......... .0...........27.......20...........7

This is doable. Say Clem has two starts in which he is shelled, out of the game before the third inning and another rough outing where he hits the showers before the fifth inning is complete. This means that 4 more of his starts fail to qualify as quality starts, going less than six innings or hitting six innings but giving up 4 or more runs.

Twenty-seven times he goes more than 6 innings and yioelds 3 runs or fewer. We know he has no shutouts, because he has no complete games. He has probably been scored on at least once in every start. Otherwise, he may have argued for, or been encouraged to go for the shutout.

All in all, Clem yields 87 runs, 81 earned, for an ERA of 3.40. In a lteam scoring atmosphere of 4.75 runs per game behind him, league atmosphere of 4.75, he'd be expected to win @ 18 games. Even if we toss in the unearned runs, he's a 17 game winner.



If 6 inning wins aren’t a good enough barometer of a pitcher’s success, then that seems to be a flaw with the stat, not the pitcher earning that stat. When people bash these pitchers and their six inning wins, they are bashing the win stat, but then those same people often turn around and extol the virtues of high win totals, even though they have already implicitly acknowledged the arbitrary nature of what it takes to get a win.

That takes a bit of sorting out. First, it's not a bash of six inning wins, per se; but is is a knock on pitchers with 5+ inning wins plus a 6+ inning maximum expectation. A guy with 34 starts, 0 CG, 187.0 IP and a 3.25 ERA may yield 70 runs [17 fewer than Clem]; but he qualifies for fewer wins because he's not around long enough. He may go 11-4 or 14-5; or with only meager luck go 10-9. To argue his effectiveness while he is in the game as a case for wins being discounted is absurd. It throws out the work factor completely. It also begs for some retroactive credit for fine work early BUT when the game outcome was not decided.

This argument also ignores the fact that low innings starters can't be considered "aces" simply because the less a starter keeps his team effectively in the game, the more dependent the team becomes on second, third and fourth line pitchers. They blow it, there should be no retroactive award for wilted starters.


(Also, why use high win totals as a proxy for durability or ability to go deep into games when we have very straightforward stats established just to measure those things, like, I dunno, innings pitched. And finally, the same people who bash the six-inning, three-run quality start, are the same people to assign positive value of the pitcher who goes eight and gives up five, while earning a 6-5 win.

I, for one, don't bash the 3 run/6 inning QC on its own face; but let's face it: It converts to a 4.50 ERA which is not much better than league average. Quality might be better replaced with "Qualified Start."


The lore and the romance of the game is one thing, precisely measuring the performance of those who play it is something different. It’s important that people remember this.

Lore and romance have nothing to do with diminished expectation.

[digglahhh]

I’m not really sure what most of your responses to my post mean, or how much they actually have to do with the value of a win, to be honest. Wins are a team stat; there need be no further discussion.

The tortured logic of those who extol the virtue of the win as anything but a proxy for any number of skills for which there exist myriad better, more precise, and more independent means of measurement is not only arcane, shortsighted and shallow, but often internally inconsistent to boot. That’s my point.

Wins have value in measuring the the performance of a pitcher, as does a carnival act in determining the weight of the next individual who walks past. But, those measurements are worthless in comparison to more finely calibrated tools, and those who take account of additional context.
One thing I want to clear up, as I may have invited misinterpretation with my post, is that I don’t consider the “quality start” as any sort of wonderful stat. And, the fact that it implies a value judgment pretty much invalidates it from even being a stat in the first place. I’ll take the components and analyze them on my own just fine, thank you. Something being considered a “quality start” doesn’t give me any new information, it’s just a way of classifying a couple of pieces of already given info about what a pitcher did. As for the question of whether 6 IP and 3 ER was a quality start, I dunno; are we in the Astrodome 1967 or in Mile High in 1998?

However, just for the record, if a pitcher goes out and gives you 34, 6-inning, 3-earned run starts, that’s more than 200 innings of roughly MLB-average pitching (depending on the environment), and that is actually quite valuable.

At the end of the day, I don’t think the “W” column has much value at all for a pitcher. It doesn’t tell me anything that many, many other numbers don’t tell me more accurately.

[dl4060]

I don't really see wins as having much value. I do see situational pitching as being relevant, maybe a pitcher being more aggressive when he is up by several runs. At that point you don't want to walk anyone, you want to make sure you don't give them anything. Someone who tries to paint the corners when the score is close might not do so with a solid lead, which could lead to a meaningless run or two. Not all runs have the same level of value, and pitchers should recognize that.

One might risk walking someone to avoid an extra base hit in a close game, while with a big lead one might go after hitters more, since a solo hr does not change the game as much. While this could lead to a slightly higher ERA, I doubt that it actually would. Of course, a pitcher who is too careful will also have big innings, so I'm not sure if any of this is really relevant.

Having said that, I don't see scenarios like the above having much to do with serious analysis. I see wins as a team stat, not an individual one.

I do give won-loss a bit of thought, but really very little. I see it as slightly relevant in the context of the pitcher's team. Someone with a .500 record and a 150 era+ on a .400 team, is better than someone with a .650 record and a 120 era+ on a .550 team.

[GordonGecko]

I think wins and losses can say something about a player, but alone isn't very meaningful. You can have an amazing player on a crappy team with a ton of losses or a mediocre player on an amazing team with lots of wins and that doesn't reflect anything about that player.

All things being equal, if there was a way to zero out team quality, relative wins and losses would say something about a player's ability to get the job done. Some pitchers will give up just enough runs to win with a high ERA, but will always give you what your team needs. That's where it means something

[curveball]

I don't really see wins as having much value. I do see situational pitching as being relevant, maybe a pitcher being more aggressive when he is up by several runs. At that point you don't want to walk anyone, you want to make sure you don't give them anything. Someone who tries to paint the corners when the score is close might not do so with a solid lead, which could lead to a meaningless run or two. Not all runs have the same level of value, and pitchers should recognize that.

One might risk walking someone to avoid an extra base hit in a close game, while with a big lead one might go after hitters more, since a solo hr does not change the game as much. While this could lead to a slightly higher ERA, I doubt that it actually would. Of course, a pitcher who is too careful will also have big innings, so I'm not sure if any of this is really relevant.

Having said that, I don't see scenarios like the above having much to do with serious analysis. I see wins as a team stat, not an individual one.

I do give won-loss a bit of thought, but really very little. I see it as slightly relevant in the context of the pitcher's team. Someone with a .500 record and a 150 era+ on a .400 team, is better than someone with a .650 record and a 120 era+ on a .550 team.


The pitcher with the era+ of 150 is better than the pitcher with the era+ of 120 as long as there isn't a huge discrepancy innings pitched.

Team won/loss records are really irrelevant, as what really matters is how many runs a team scores for a particular pitcher in each of his starts. Of course, one would expect a team with a high winning percentage to score more runs, and they do, but they don't always distribute this run support evenly among all the pitchers in the starting rotation. Run support can and will vary greatly, even among starting pitchers on the same team.

Wins are a team stat. Wins that are assigned to a starting pitcher are virtually useless. Someone decided to arbitrarily reward pitcher's with wins and saves based on their guidelines. Wins could have easily been awarded to the pitcher who pitched the most innings for the winning team. A save could have simply been given to the pitcher who registered the last out for the winning team.

You don't even have to pitch well to get a win or a save, and that itself should be a sign as to how misleading they are.

[leewileyfan]

I’m not really sure what most of your responses to my post mean, or how much they actually have to do with the value of a win, to be honest. Wins are a team stat; there need be no further discussion.

First, earlier in this thread I did post my belief that pitcher W-L records and morning starting lines were initially conceived as media promotion exercises to put pitchers on the marquee as draws. It fit well with the U.S. mano a mano shoot-out psychology and drew the attention of fans.

Second, the promotional conception had value as a practical recognition for pitcher efforts in an age when starters were expected to complete most of their games.

Third, in the absence of enclosed domes, something had to define a game qualified to go in the scorebook as official. Half-a-game [4.5 innings] seemed both practical and undemanding. If a Win is assumed to be a team stat, why shouldn't it be applied as well to a pitcher who throws an official game?

Fourth, batters have all sorts of neon stats to highlight their accomplishments: BA, OB%, SLG; HR, RC; RS; RBI; yet the guy in the middle of the diamond, the only one presumed to be involved in every batter sequence, is to be begrudged a Win. A pitcher can feasibly go 27-2 and not even make it to the post-season [don't hold your breath]. His 27 wins are to his credit. His team's 85 wins may as easily not cut the mustard.

Fifth, any fan realizes that this argument conveniently omits 2 key adjectives: pitcher and team. One adjective defines the contribution credited to individual performers; the other is the aggregated sum of team performance. Where is the harm?


The tortured logic of those who extol the virtue of the win as anything but a proxy for any number of skills for which there exist myriad better, more precise, and more independent means of measurement is not only arcane, shortsighted and shallow, but often internally inconsistent to boot. That’s my point.

There is not the slightest hint of [I]"tortured logic" in arguing for utter simplicity as a starting point. I can glance at two pitchers' records, one 27-2; the other 11-15. If All that tells me immediately is that the first had a good year. If I know baseball, I will not allow myself to consider the other guy a bum. I have to dig deeper. But I do have a decent start [pun intended].


Wins have value in measuring the the performance of a pitcher, as does a carnival act in determining the weight of the next individual who walks past. But, those measurements are worthless in comparison to more finely calibrated tools, and those who take account of additional context.

Disagree. Wins are far less randomly generated than guessed weight of carnival passers-by. You have also alluded to more nuanced stats that are more telling. However, like pieces in a jigsaw puzzle, they give a piece of the bigger picture as their very descriptor nuanced implies. Add them up and you're in a position to attack the relative merits of two pitchers [27-2; 11-15] and their W-L records. At this point you identify modifiers to cold summaries. You do not demolish the essence of the W-L stat itself.



One thing I want to clear up, as I may have invited misinterpretation with my post, is that I don’t consider the “quality start” as any sort of wonderful stat. And, the fact that it implies a value judgment pretty much invalidates it from even being a stat in the first place. I’ll take the components and analyze them on my own just fine, thank you. Something being considered a “quality start” doesn’t give me any new information, it’s just a way of classifying a couple of pieces of already given info about what a pitcher did. As for the question of whether 6 IP and 3 ER was a quality start, I dunno; are we in the Astrodome 1967 or in Mile High in 1998?

Agreed. The Quality Start is, to me, of little value. Lasting 2/3rds of a game @ a league average runs allowed is a questionable achievement.


However, just for the record, if a pitcher goes out and gives you 34, 6-inning, 3-earned run starts, that’s more than 200 innings of roughly MLB-average pitching (depending on the environment), and that is actually quite valuable.

OK. He's 204.00 IP, no CG, no ShO and with a 4.50 ERA. If this is "quite valuable," I am guessing that he's number 1, 2, or 3 in the rotation. This suggests a staff where at least one guy in the rotation has a worse record. It also suggests that if the guy under our scrutiny is "quite valuable," any starter ahead of him may be better, but not too much better.

Is this club a contender?


At the end of the day, I don’t think the “W” column has much value at all for a pitcher. It doesn’t tell me anything that many, many other numbers don’t tell me more accurately.

For relievers, especially closers, I couldn't agree more. For starters, that's another question.

[dl4060]

The pitcher with the era+ of 150 is better than the pitcher with the era+ of 120 as long as there isn't a huge discrepancy innings pitched.

Yep. I forgot to mention innings. Kind of a big deal.:D Obviously they need similar numbers of innings to make it a good comparison.

[Los Bravos]

I can glance at two pitchers' records, one 27-2; the other 11-15. If All that tells me immediately is that the first had a good year. If I know baseball, I will not allow myself to consider the other guy a bum. I have to dig deeper. But I do have a decent start [pun intended].

Wins are far less randomly generated than guessed weight of carnival passers-by. You have also alluded to more nuanced stats that are more telling. However, like pieces in a jigsaw puzzle, they give a piece of the bigger picture as their very descriptor nuanced implies. Add them up and you're in a position to attack the relative merits of two pitchers [27-2; 11-15] and their W-L records. At this point you identify modifiers to cold summaries. You do not demolish the essence of the W-L stat itself.That's pretty much my opinion, as well.

I also think that the larger career sample size you look at, the more representative the W-L record becomes. There are outliers, but you don't usually get mediocre or merely good, pitchers compiling really excellent records (in terms of either raw win totals or winning percentage) over the long haul.

Similarly, absent a few examples like Bert Blyleven, you don't get outstanding starters compiling underwhelming numbers in those categories, absent some major issue (like Ryan's congenital wildness.)

The stat isn't the Alpha and Omega that it's generally been thought of over time, but this overcorrection has really gone too far. You need to look deeper, but you have to start somewhere, and the basic stats are as good a place as any.

If you want a metaphor, it's like diagnosing a patient initially by noting that he looks healthy and isn't canary yellow or vomiting profusely. He's probably healthy but it would be a good idea to run some tests and be sure.

[digglahhh]

Lee & Bravos,

I don't necessarily disagree entirely. I guess we're just talking about a different bar of usefulness. Yes, "Wins" are better than nothing, and they are a starting place that can give us in an inclination as to the quality of a pitcher. But, we don't peel off a pitcher's stats, one-by-one, like were poking open an advent calendar or something. There is a ton of information out there, usually consolidated in one or a very few places, so I don't have much use for the "win." Maybe I overstated things by saying "useless" or "worthless," however, I only see the stat as being useful, or worthy in the absence of many, many other stats.

[STLCards2]

Lee & Bravos,

I don't necessarily disagree entirely. I guess we're just talking about a different bar of usefulness. Yes, "Wins" are better than nothing, and they are a starting place that can give us in an inclination as to the quality of a pitcher. But, we don't peel off a pitcher's stats, one-by-one, like were poking open an advent calendar or something. There is a ton of information out there, usually consolidated in one or a very few places, so I don't have much use for the "win." Maybe I overstated things by saying "useless" or "worthless," however, I only see the stat as being useful, or worthy in the absence of many, many other stats.

Correct - givin no other information, you could make a decent all-time career list off of pitcher's W-L record. However, we do have other information...lots of information. With that information we can come up with very good all-time career lists.

[brett]

Correct - givin no other information, you could make a decent all-time career list off of pitcher's W-L record. However, we do have other information...lots of information. With that information we can come up with very good all-time career lists.

I've got a question for you. We can take a pitcher who has a 16-8 record with a 200 ERA+ like Greinke and we can make a good estimate that an average pitcher might have gone 7-16 with the same run support, so he was still like a 21-4 in terms of value above .500 for his team.

But what about a season like Nolan Ryan had in '87? Can a guy who went 8-16 actually have been an all-star level pitcher for his team?

Even if an average pitcher would have gone 6-18 with his run support, he's only 2 wins above that, like say 14-10 worth in context.

I know I'm oversimplifying the math but I'd like people to compare the following sets of seasons:

Ryan '87: 8-16 142 ERA+
Clemens '05: 13-8 226 ERA+
Johnson '04: 16-14 177 ERA+
Steib '85: 14-13 172 ERA+



Versus
Clemens '01: 20-3 128 ERA+
Hoyt '83: 24-10 115 ERA+
Welch '90: 27-6 126 ERA+
Jack Morris '84: 19-11 109 ERA+

Group 1 posted about a 180 ERA+ and 13-13 average record.

Group 2 posted about a 120 ERA+ and 22-7 record.

Do we honestly feel that if we take the group 1 seasons, we don't just get a bunch of 22-7 records, but actually get near perfect records? In fact my pythag adjustment says that group 1 would have gone on average 31 and -2 with the run support of group 2.

[Ubiquitous]

Nolan in 1987 had stretches where he wasn't very good and he got the loss and then he had stretches where he was very good and didn't get the decision.

He pitched 9 games in which he gave up only 1 earned run. Two of those went less than 5. But of the 7 that went more than that he had a record of 2-2 with 3 no decisions. Of the 10 games in which he gave up only 2 earned runs he only won one of them and lost 5 games. So that is 3-7 with 7 no decisions while giving up 1 or 2 runs and pitching 5 or more innings. Is that a typical line for a starter with an average offense? I don't know but I kind of doubt it. He also goes 1-7 in any games in which he gives up more than 2 runs and he gets 2 no decisions. Is that typical? Again I kind of doubt it.

[leewileyfan]

Brett's introduction of several pitchers and selected seasons got me to working on a formula [still under construction] that attempts to interpret pitcher value in virtual [earned] W-L by way of WHIP modified by a K-based factor.

Here are some of the seasons & pitchers, their W-L records and the formula based shoulda-been recoreds, FWIW:

Pitcher............Season...........Record........ ...Earned W-L

Clemens ....... 2005............ 13-8 .............. 16-5

*Oswalt....... 2005............ 20-12............. 21-11

*Oswalt, not on Brett's list, was a Clemens teammate in 2005 and the ace of the Houston staff.

Welch......... 1990............ 27-6............... 19-14

Clemens...... 2001............ 20-3............... 15-8

Ryan.......... 1987............ 8-16.............. 15-9

ERA+, in my humble opinion, isolates a pitcher's record much to far from the larger dynamics of the game. Although I haven't yet proofed my work for 2005 [still sweating through the 1990s], nores tell me that both Clemens and Oswalt had excellent defensive support [@ 34-36 runs better than League average] and were -28 season batting runs overall. The HOU record of 89 wins, finishing in 2nd place supports this.

Ryan's 1987 final numbers in W-L were the biggest screwing of the lot.

[Los Bravos]

I propose a 6 month moratorium on referencing either Nolan's '87 or Welch's '90.

I think we all realize that it's a potentially inaccurate measure of worth and effectiveness, especially the smaller the sample size you use. As I said, it's merely a starting point. Anyone who cares about figuring out the truth will want to look deeper.

I'll offer a semantic compromise. I had an epiphany about this this morning, namely that I look at wins and a good W-L record, especially over the course of several years and a career, as a matter of accomplishment.

Everyone here taking the position that they're not that big a deal has stressed their inadequacy as a measuring tool and the fact that there are better ways of analyzing what a guy brings to the mound on a daily basis. I don't disagree. I just don't like seeing people who have stellar records denigrated because they had good offense and/or defense behind them. I don't see that as anything shameful and I don't see having to struggle without them as any sort of virtue. I can guarantee that Bert Blyleven would have preferred to have more of both throughout his career.

Finally, not to sound too cornpone about it, but the fact remains that these things often take care of themselves. Clemens appears on both lists. brett references Randy Johnson's 2004, but I could just as easily point to his 2006 season, where he wound up with 17 wins with a truly lousy 5.00 ERA (that's 90 ERA+) Baseball-Reference.com has his neutralized record that year as 9-13. They have him at 12-12 in '05, when he actually won another 17 games.

In fact, his two year tenure in the Bronx was a case study in how you can benefit from being on a great team. His actual record was 34-19 which he managed with a 4.37 ERA, which translates to 100 ERA+. Randy Johnson won 34 games in two years as (essentially) a replacement level pitcher. Having said that, does anyone think he's not really a 300 game winner, with all that entails? I hate his guts but I recognize that this bloated record is balanced out by other times he was hugely dominant and didn't get a proper reward for it.

In the end, Bob Welch didn't ride that inflated total into Cooperstown. It was a one year aberration for a talented major league pitcher. Under even the harshest assessment of that year, I think he finishes high in the Cy Young balloting, coming up short to Clemens. Similarly, Hoyt never duplicated that outlier year.

[STLCards2]

Group 1 posted about a 180 ERA+ and 13-13 average record.

Group 2 posted about a 120 ERA+ and 22-7 record.

Do we honestly feel that if we take the group 1 seasons, we don't just get a bunch of 22-7 records, but actually get near perfect records? In fact my pythag adjustment says that group 1 would have gone on average 31 and -2 with the run support of group 2.

I would bet it has to do with the distribution of their earned runs. Also, some people do not like to talk about it because it is hard to statistically prove "why"- but there have been some pitchers who do consistently pitch better or worse in close games compared to blow-outs. Bert Blyleven was relatively horrible in close games compared to blowouts (which shows that all of his record vs. expected record was not lack of offensive support), Not sure about Ryan. Maddux and Glavine, for example, pitched much worse in blowouts than in close games.

And the two poster boys for this skill - Hunter and Morris, are both about the same either way, contrary to many peoples' hopes.

[curveball]

I would bet it has to do with the distribution of their earned runs. Also, some people do not like to talk about it because it is hard to statistically prove "why"- but there have been some pitchers who do consistently pitch better or worse in close games compared to blow-outs. Bert Blyleven was relatively horrible in close games compared to blowouts (which shows that all of his record vs. expected record was not lack of offensive support), Not sure about Ryan. Maddux and Glavine, for example, pitched much worse in blowouts than in close games.

And the two poster boys for this skill - Hunter and Morris, are both about the same either way, contrary to many peoples' hopes.


I am not sure that you can argue that Blyleven's actual record was not really because of lack of offensive support.

Baseball-reference has something called cheap wins, tough losses, losses saved, and wins lost.

Blyleven had 35 cheap wins, which are when you pitch less than 6 innings, give up more than 3 earned runs, yet still record the win.

He had 99 tough losses, which are games that a pitcher throws a quality start but gets the loss.

He had 43 wins lost by the bullpen.

He had 58 losses saved by his team coming back to tie or take a lead when he was in a position for a loss.

If you subtract his 35 cheap wins from his 99 tough losses, then he is at negative 64 wins. He gains 15 wins when you subtract his losses saved from his wins lost. This leaves him at negative 49 wins for his career. I am pretty sure that this would be among the worst totals in history for a starting pitcher, so his actual record definitely should have been better than it was.

I added Andy Pettite, and Kenny Rogers, to illustrate pitchers who have very good won/loss records, but not very good earned run averages.

Here are some other pitchers,

BLYLEVEN -49
KOOSMAN -49
RYAN -44
PERRY -43
JENKINS -26
NIEKRO -25
CLEMENS -20
GIBSON -19
CARLTON -13
MADDUX -13
P MARTINEZ -11
PALMER -8
SCHILLING -5
SMOLTZ -1
R JOHNSON 0
MUSSINA 0
MORRIS +8
GLAVINE +9
MARICHAL +11
PETTITE +20
ROGERS +33

[Los Bravos]

And the two poster boys for this skill - Hunter and Morris, are both about the same either way, contrary to many peoples' hopes.You mean "Pitching to the score"? I read Sheehan's examination of Morris' record that was reposted here a few months ago. It was pretty convincing. (Keith Law referenced it in one of his online chats after his Cy Young votes this year, as well.)

Even so, I think that exists, but it's more of a mindset or approach to pitch selection than something that is easily quantifiable, at least by combing through the numbers. I think it correlates less to the score than to the overall approach of pitching to contact and trying to avoid walks at nearly all costs, including solo homers. Gibson's approach late in Game 7 of the '64 Series comes to mind.

[STLCards2]

I am not sure that you can argue that Blyleven's actual record was not really because of lack of offensive support.



I said not "all" due to lack of run support. Not "really."

[leewileyfan]

Rating & evaluating pitcher performance is always both interesting & challenging.
This response has to do with starting pitchers only, and then, only those whose careers lay dominantly within the 1901-2009 time span.

The numbers below attempt to display starters from two perspectives:

1. Legacy, Career, "Greatness Points" accumulated points, which attempts to take the relative adjusted ERA for each pitcher and apply that to a longevity factor. Therefore, guys with extended, effective careers get special notice IF they generally maintained a kind of dominance over their extended numbers of years.

2. ERA, from whatever era, adjusted to all eras, for an isolated one-on-one comparison to any/all other pitchers under consideration; like a barroom debate along the lines of "What if Pitcher N faced off against Pitcher Q, each on one of his best days?

This whole post was prompted by STL's mention of Jack Morris and Catfish Hunter, whom I get to at the end.

Pitcher.........................Era-Adjusted ERA..............Legacy Points

W. Johnson..................... 2.25.............................1,479.2
Clemens......................... 2.54..............................1,072.7
Maddux.......................... 2.55..............................1,062.3
Alexander....................... 2.81.............................. 976.1
Grove............................ 2.30............................. 964.1
Mathewson..................... 2.77.............................. 918.0
Seaver........................... 2.71.............................. 908.4
Hubbell........................... 2.34.............................. 861.4
*Johnson, R................... 2.78................................748.2*
P. Martinez...................... 2.12.............................. 746.2


*Randy Johnson's omission on original list now corrected.

*C. Young........................ 2.67............................. 740.2

*[Cy Young's numbers are reduced because all seasons prior to and through 1900 are not included.

Spahn............................ 3.25............................. 731.1
E. Walsh......................... 2.41............................. 686.9
Smoltz........................... 2.78............................ 650.3
Feller............................. 3.02............................. 630.1
Ryan.............................. 3.51............................. 594.4
Vance............................ 2.75............................. 577.4
Lyons............................. 3.02............................. 556.1
Schilling.......................... 3.02............................. 537.4
Ruffing............................ 3.40............................. 533.1
Mussina.......................... 3.16............................. 530.2
**new addition: Luque, Dolf...... 3.04................... 521.5
K. Brown......................... 3.03............................. 521.4
Derringer........................ 3.15.............................. 505.9
Joss.............................. 2.55............................. 504.0

IMHO, any starting pitcher accumulating >500 Legacy/Career "Points," should be considered a no-brainer inductee into the HoF. In fact, given the group that follows, I'd argue for any pitcher with >400 such "points" should also be in on sheer volume of accomplishment, maintained competitive superiority[regardless of career W-L record, and having longevity with such performance.

E. "W" Ford................... 3.08............................. 498.8
Glavine......................... 3.49.............................. 496.5
M. Brown...................... 3.12............................. 488.1
"Dutch" Leonard............. 3.14............................ 487.1
G. Perry........................ 3.70............................. 477.0
Gibson.......................... 3.40............................. 474.7
Warneke....................... 2.97............................. 472.1
Carlton......................... 3.69............................. 469.5
Palmer.......................... 3.45............................. 461.4
Roberts......................... 3.64............................. 446.3
Gomez.......................... 2.91.............................. 443.0
Wilhelm......................... 2.82.............................. 420.0
Jenkins......................... 3.64............................... 427.6
Blyleven....................... 3.73............................... 426.1
Walters......................... 3.28.............................. 421.2
*Eckersley..................... 3.37............................... 413.5

*[Eckersley, renowned as a reliever, is here because he was, for years, an effective starter ........... no reliever, isolated in that role, has >400 "points"]

Marichal........................ 3.46............................... 406.5
C. Mays........................ 3.30............................... 402.9

Many HoF are not on these lists; and several non-HoF are. I don't believe a well-considered argument can be made against any of these.

Now, as to Jack Morris and Jim "Catfish" Hunter. Neither rates > 300 or even > 200 "points in this evaluation metric. This does not condemn either or both as bad pitchers; but I believe it makes an honest evaluation of their performance against the standards set for the metric.

"Catfish" Hunter............. 4.19.............................. 119.0
Jack Morris................... 4.16............................. 143.9

For context, here are other starters within this range:

Blackwell [108.24; 3.76]; Hershiser [124.0; 4.14]; Bender [159.9; 4.02]; Koosman [124.7; 4.21]; Trucks [166.31; 3.94]; Chance [151.4; 3.87]; Shantz [126.2; 3.92]; P. Niekro [146.1; 4.26]; John [122.4; 4.27]; C. Finley [118.0; 4.13].

Sometimes, excellent pitchers have gotten buried in smaller media franchise cities; have ground out workhorse efficiencies over long careers with feckless clubs. I believe this evaluation method shines the light on some of the most deserving forgotten men.

[brett]

Rating & evaluating pitcher performance is always both interesting & challenging.


Is this a complete list? I don't see Randy Johnson.

[leewileyfan]

Is this a complete list? I don't see Randy Johnson.

You have but to ask, and it shall be corrected. Randy Johnson [748.2; 2.78] rates just ahead of Pedro Martinez. Please see original listing where RJ is properly inserted. Thanks for the catch.

[brett]

You have but to ask, and it shall be corrected. Randy Johnson [748.2; 2.78] rates just ahead of Pedro Martinez. Please see original listing where RJ is properly inserted. Thanks for the catch.

I asked because the ranking is very close to a typical top 12-13 around here. At one point, my list matched your leaders very closely. I had

Walter Johnson
Clemens
Young
Maddux
Alexander
Grove
Mathewson
Seaver
Randy Johnson
Pedro Martinez
Hubbell
Feller (with war credit)
Spahn
Nichols (19th century addition)

The major differences then begin to appear. I have come to put longer career guys like Carlton, Perry and Jenkins ahead of shorter dominance guys like Joss and M. Brown.

You do seem to be picking up on Brown's good defenses helping him. Before I saw any defensive adjustments, I had him in my 14-20 range and similar to Hubbell, Gibson.

Joss supposedly had very good defenses as well though and I wonder if he doesn't benefit from not having a decline period in your ratings.

You have Mussina, Smoltz, K. Brown and Schilling high which is good, but I just don't see Niekro or Gibson that low. Perhaps Niekro really does outdo his WHIP estimates somehow being a knuckler and able to control extra base hits somewhat.

Where do you have Tiant? Matt's defensive adjustments say he was much better than his ERA+ indicates.

Gomez would be a suprise for some to see as a near lock hall of famer. I don't mind him in the hall but he is a controversial one. What makes him better in your system than Dave Stieb or Brett Saberhagen?

Where does Koufax fit in?

[STLCards2]

He had 43 wins lost by the bullpen.

He had 58 losses saved by his team coming back to tie or take a lead when he was in a position for a loss.



You can get a "loss saved" by being behind 1-0 when you leave after the top of the 7th, and the offense scoring 2 in the bottom of the 7th to win the game. These kind s of games happened a lot for the greats. Without knowing the distribution of runs allowed for their "losses saved" or "wins lost", it is a limited (albeit interesting) stat to use. that is why it is better to use linear weights and convert them into some WAR type measure to begin with - bullpen and run support are completely weeded out.

By the way, here are the margin <> 4 tOPS+ for the pitchers mentioned:

Average = 100
Higher than 100 means pitcher pitched better during close games, lower than 100 means the pitcher pitched worse when the game was close. (and/or that a pitcher gave up more/less runs when it mattered less/more). In other words, the higher, the better.

Blyleven - 85 :eek:
Ryan - 100
Hunter - 99
Morris - 101
Glavine - 108
Maddux - 116 :eek:

Here are a couple for anybody following the recent Gibson vs. Marichal debates:

Gibson - 85 :eek:
Marichal - 104

Don't let Roy see this or we will never hear the end of it. :)

Also:

Palmer - 99
Niekro - 98
Carlton - 95


Here are a few more some may want to see:
Mussina - 103
Smoltz - 109
Schilling - 97
K. Brown - 100
Johnson - 100
Pedro - 102
Clemens- 106

Seems like most of these guys are a little better when the game is close - with some being quite a bit better. Blyleven seems to have pitched significantly better when his team was already up or down by 4 or more runs. Now how does that effect his ERA or W-L record or WAR? Not sure, but he is definitely a HOF anyway, so it isn't a make or break thing - just interesting more than anything else.

And last but not least...

Sandy Koufax - 125. If the game was close - he was even more dominating and lights out or when he had the very occasional off day to begin with, it snowballed and he was dreadful.

Anyway - just food for thought. Bottom line, run distribution can be a factor in w-l record. It is not as significant as run support, but shouldn't be ignored either.

[leewileyfan]

Where do you have Tiant?

Tiant is @ 256.57 with adjusted ERA of 3.79. This seems more than fair if one considers that 10 of his nineteen seasons were mediocre.


Gomez would be a suprise for some to see as a near lock hall of famer. I don't mind him in the hall but he is a controversial one.

I am not being arrogant here; but I do have faith in the numbers because I've done everything in my power to remove bias. Lefty Gomez pitched in one toughest and most hellacious decades in modern MLB history, worked a ton of innings [some in relief, while he was in the regualr rotation]; and despite a career obviously abbreviated by arm trouble, dominated for the most part. He belongs in the HoF without any doubt at all.


What makes him better in your system than Dave Stieb or Brett Saberhagen?

Stieb [240.1; 3.75] and Saberhagen [207.1; 3.77] were both solid and dependable pitchers for their generations; but neither had the combined relative dominance above their league medians nor the consistency of such performance over the mass of career as did Gomez.



Where does Koufax fit in?

I have Koufax @ 366.62; 3.08, largely a result of three input factors: brevity of career, arthritis; erratic and mediocre start; very light hitting atmosphere. The absence of a decline period for Koufax in fact props his numbers up. Death [a high price indeed at 31] sustained Addie Joss' numbers from a period of gradual decline.

I am not one much for insisting on a decline period for MLB players in tallying up the totality of each one's final evaluation. I take the career as presented to me. I don't hold personal career-ending personal calamity against a performer to argue he failed to experience a decline period experience by most.

[brett]

I am not one much for insisting on a decline period for MLB players in tallying up the totality of each one's final evaluation. I take the career as presented to me. I don't hold personal career-ending personal calamity against a performer to argue he failed to experience a decline period experience by most.

My question regarding this was whether your SYSTEM would favor a player if his decline period was removed from the books.

What would Seaver's rating be if he had retired after '81? Could you give his actual legacy points and what they would have been through '81?

What would Maddux have been if he'd retired after '02? Same question.

[leewileyfan]

My question regarding this was whether your SYSTEM would favor a player if his decline period was removed from the books.

The most comprehensively honest answer I can give is that it would depend on the following factors:

1. Length of the pitcher's career overall;
2. Duration of decline period to the whole of the career;
3. Change in career batting climate faced between the bulk of the pitcher's best years and years of his decline;
4. Volume of innings, decline period to the whole


What would Seaver's rating be if he had retired after '81? Could you give his actual legacy points and what they would have been through '81?

Actually, Seaver was quite effective after 1981, in '83, '84, and '85. Moreover, he ended his career facing a tougher offensive climate than he faced in his early seasons.

1982 and '86 would have had a small impact; but the volume of the good years would bury those two seasons and their innings.

Estimate: Seaver career overall, as stated: 908.4; 2.71
Estimate: Seaver, retired after 1981 season: 2.69; 910


What would Maddux have been if he'd retired after '02? Same question.

Maddux is similar to Seaver on all counts: effective in '03 through '07.

Maddux, career, as stated: 1062.3; 2.55
Maddux, Estimated, retired after 2002: 1066.0; 2.52

Both would maintain standing in career valuations.

[brett]

Actually, Seaver was quite effective after 1981, in '83, '84, and '85. Moreover, he ended his career facing a tougher offensive climate than he faced in his early seasons.

1982 and '86 would have had a small impact; but the volume of the good years would bury those two seasons and their innings.

Estimate: Seaver career overall, as stated: 908.4; 2.71
Estimate: Seaver, retired after 1981 season: 2.69; 910

Maddux is similar to Seaver on all counts: effective in '03 through '07.

Maddux, career, as stated: 1062.3; 2.55
Maddux, Estimated, retired after 2002: 1066.0; 2.52

Both would maintain standing in career valuations.

See, that's my beef. Seaver and Maddux were (fairly) unique in their ability to produce value in their last 5 seasons. They DID have to adapt-which I consider to be a skill-to having less velocity and lower K rates. You give them nothing in their overall rating for posting quality years when most guys would have been retired. And while they may have been #3 level starters, if you drop their production you get someone with a 4.5 ERA taking their place.

I just think that there have been some pitchers who lose their top stuff in the early 30s, and some who adapt and some like Clemens, Ryan and Randy Johnson who still have it, but the adaptation ability is something that I value.

[leewileyfan]

See, that's my beef. Seaver and Maddux were (fairly) unique in their ability to produce value in their last 5 seasons. They DID have to adapt-which I consider to be a skill-to having less velocity and lower K rates. You give them nothing in their overall rating for posting quality years when most guys would have been retired.

Honest question [and to cite a classic tv commercial]: "Where's the Beef???

What would you have an evaluator do - fudge some numbers so a 38 year old pitcher, one a 20 game winner and Cy Young voting prospect gets a boost up from his actual on-field performance because most guys his age have gone fishing?

1. MLB players at all positions reach a stage when they have to make an informed decision, one might hope is not primarily motivated by an economic need to hang onto a job:

-How do I stand career-wise at this moment? Am I 67 base hits away from 1,000? 2,000, 3,000, 4,000? Each has its own frame of reference. Each has its own context. The guy 67 hits away from 1,000 may also be 6 months short of qualifying for a ten-year MLB active player pension.

-For a pitcher @ 40 years of age who went 5-17 last season, with an ERA of 5.60, why is he pondering coming back next spring? Does he have 297 career wins so far? Or is he, too, at career 117-101 just a few months short of the 10 year career mark?

-A guy with a lifetime BA of .2997, now 37 and who has squandered his earnings and invested badly, may be dying to go out a .300 hitter; but another year or two as a marginal regular or oft-used utility man may give him two more seasons @ $650,000 each even though he'll drop to .296 in the process.

It's all right there in the career numbers. That is the player's legacy. Whatever motivation lies behind those numbers, all I can do is make a best case, fair evaluation; and my metric tries to make comps across generations with as little bias as possible.

Maddux and Seaver are not penalized by my metric any more than their performance would demand. They are sustained because they adapted.


And while they may have been #3 level starters, if you drop their production you get someone with a 4.5 ERA taking their place.

That begs the question, big-time. The kid replacing them may be the next Pedro Martinez or the next Hoyt Wilhelm.


I just think that there have been some pitchers who lose their top stuff in the early 30s, and some who adapt and some like Clemens, Ryan and Randy Johnson who still have it, but the adaptation ability is something that I value.

In all fairness, you speak of adaptability as something that can be willed into being, very slightly suggesting that those who don't adapt could, if they really wanted to.

Of course, MLB, like all fields has its share of loafers, louts, malcontents and rock-headed one-note guys who either refuse to adapt, or other guys who simply can't adaptdue to the nature of their setback[s].

Bottom-line, for me: career=legacy.

Post Script:

Anyone want to talk about adaptability, take the case of Emil "Dutch" Leonard, who broke in at 24 with BRO [134]. After going 14-11 the next year, he suffered an arm injury; and he struggled @ 2-9 the next year, virtually going out of MLB to rehabilitate and learn to pitch again. He came back after being out '35-'37, at 29 now having learned and mastered the knuckleball. Toiling with losing clubs, he made a full comeback. Then he missed almost all of 1942 with a broken leg.

He came back again in 1943 and had a 20+ year MLB career, winning 191 games, often with cellar dwellers. He wrapped it up as a superb reliever with the Cubs.

Difference between Leonard and other Knucklers: he mixed it up with fastballs and sinkers, with excellent control. I was lucky enough to see Leonard several times when he was with WAS; and he threw everything with a fluid motion - but not the easy backyard game-of-catch motion of a Wilhelm or a Wakefield. His knuckler was thrown with the same motion ast his fastball - doubling as a knee-locking change-up.

He adapted.

[Los Bravos]

A few minor things I want to add. First, this has been a terrific discussion. This is a great example of what I come here for. I'm particularly fascinated by the runs distribution post Cards put up. That Koufax number doesn't surprise me one bit.

Secondly, (and I don't want to start a tangent about PEDs, but...) I value adaptability, too. It's the main reason that I devalue the second half of Clemens' career. It seems clear to me that his motivation was mainly because of his being unwilling to let go of his self image as a power pitcher. He hit the same wall so many others before him have hit and rather than learn to climb over it, he dug a tunnel under it instead.

[brett]

Nolan in 1987 had stretches where he wasn't very good and he got the loss and then he had stretches where he was very good and didn't get the decision.



In other words his W-L record is worth looking at in evaluation.

[STLCards2]

A few minor things I want to add. First, this has been a terrific discussion. This is a great example of what I come here for. I'm particularly fascinated by the runs distribution post Cards put up. That Koufax number doesn't surprise me one bit. He had to be that fine with the offense he was working with.

Secondly, (and I don't want to start a tangent about PEDs, but...) I value adaptability, too. It's the main reason that I devalue the second half of Clemens' career. It seems clear to me that his motivation was mainly because of his being unwilling to let go of his self image as a power pitcher. He hit the same wall so many others before him have hit and rather than learn to climb over it, he dug a tunnel under it instead.

I have always wondered about Clemens' Yankee years. Popular viewpoint says that Clemens started using in 1997 or 1998, which if one likes to look at statistical improvement would make sense (he was still great from 1994-1996, but not quite as much as 1986-1993 or 1997-1998), but the worst (relatively, of course) period of his career came as a Yankee after he was supposedly several years into usage. His numbers did not take off into absurd-land again until he went to Houston.

We have some options here:

1. He stopped using when he got to N.Y., then started again in Houston (seems a little problematic)
2. Even on roids, age caught up with him in NY and his performance suffered, but the NL was so much worse than the AL, he dominated despite age (too many ifs - not likely at all)
3. Clemens was never on steroids or any other performance enhancer (maybe more likely than #1 or #2 - but I still doubt it)
4. We have no idea how much impact any PED can have with any given player. Clemens had a unique experience with PEDS, and looking at career paths to determine usage is a shaky premise since there are many, many PED and non-PED careers that do not follow expected trends. Maybe it is possible that for some players, PEDS do not impact production at all?

Am I missing any possibilities for his strange career shape?


Also - looking at Clemen's pitch repertoire, he increased his off-speed pitches (which of course decreased his fastball usage) in Houston, at least. That doesn't seem like the type who refused to lose a power-pitcher image. In fact, his splitter/sinker in Houston became a primary pitch (and one of his most successful.)

[Ubiquitous]

In other words his W-L record is worth looking at in evaluation.

No, it means his w-l record is largely immaterial

[Ubiquitous]

A few minor things I want to add. First, this has been a terrific discussion. This is a great example of what I come here for. I'm particularly fascinated by the runs distribution post Cards put up. That Koufax number doesn't surprise me one bit. He had to be that fine with the offense he was working with.



Koufax had one of the league's best offenses behind him. It wasn't until his last year or so that they started to slip from the top. The Dodgers played at Dodger Stadium and it did much to hide the strength of their offense while showcasing their pitching.

[leewileyfan]

Just having fun with this & from the perspective that pitcher evaluation methods can become prismatic, I decided to take the "Fame Points" in my post and divide it by career IP to see what pitchers might rise to the surface [looking for surprises]. Only pitchers with .2 pts/IP are listed, with no rounding up for .19_ guys.

For active pitchers, I believe I updated all of them through 2009; but certainlt through 2008, these were their quotients: [Points/IP]

Starters....................................Reliev ers

Martinez....... .2639..................Rivera........... .3261
W. Johnson... .2501..................Henke........... .2659
Santana....... .2457..................Hoffman......... .2611
Grove........... .2447.................Cordero......... .2535
Hubbell......... .2359..................Foulke.......... .2364
Walsh........... .2307.................Percival......... .2305
Clemens........ .2182.................Maltzberger.... .2304
Maddux......... .2169..................Wetteland...... .2228
Joss............. .2166..................Benitez.......... .2178
Young.......... .2037..................Beggs........... .2123
Halladay....... .2020..................Reardon......... .2032
Oswalt......... .2017..................Gagne........... .2027

In little tangential searches like this, not restricted by career durations or reputations, guys otherwise forgotten rise up for a second look.

In this instance, it blew my mind that I had overlooked Tom Henke in my relief pitcher evaluations. Joe Beggs in the '40s was a guy known to be effective in relief, but at a time when bullpen pitchers were far from a specialization. Good to see him show up as well.

As for Gordon Maltzberger, he broke in at age 30 and then lost a year+ to WW II; so he had little time to shine. He was unique for his time in being used exclusively in relief. In the brief time he had to shine in that role, he did.

[brett]

No, it means his w-l record is largely immaterial

You went about demonstrating that his record WAS lower than would be expected based on normal saber assessment because of irregularities in his performance. Immaterial if you have all the splits to look at, but if you take a pitcher's raw pitching performance, and adjust it for defense and run support and his record is STILL better or worse than expected, the discrepancy in his record is showing something about his pitching, in this case, consistency or lack of consistency.

[Ubiquitous]

You went about demonstrating that his record WAS lower than would be expected based on normal saber assessment because of irregularities in his performance. Immaterial if you have all the splits to look at, but if you take a pitcher's raw pitching performance, and adjust it for defense and run support and his record is STILL better or worse than expected, the discrepancy in his record is showing something about his pitching, in this case, consistency or lack of consistency.

His W-L record is still immaterial. Knowing his record reveals nothing about how Nolan pitched. It is knowing everything else that reveals something about how Nolan pitched. If you everything else and not his W-L record and then find out his record you add nothing to new to how Nolan pitched.

[Second Base Coach]

LONG POST ALERT

Ugh... the Nolan Ryan 8-16 1987 season comes up again in a thread.

In his 16 losses, he averaged five and a half innings per game.

His ERA in those losses was 4.19

I know the guy led the league in ERA and strikeouts, and ERA+ and fewest hits per game.... and so forth.

But it's not like he was losing a ton of games by a score of 2-1 after pitching eight innings.

It was not that kind of season. It jumps out at you because of the seasonal record and we really don't see a whole lot of ERA and strikeout champs (in the same season) losing twice as many games as he wins, but he varied widely in his performance.

His ERA in his wins was 1.11 for instance. And he never gave up a home run in any of the eight games he won. All 14 homers came in losses or no decisions.

In 34 starts, he made it into the seventh inning 19 times. The batting average of the balls in play after the seventh inning was .444 I just don't think he stayed in games long enough to let more good things happen. Note that he had no complete games, but again he was 40 years old at the time.

Lucky for him he only faced 92 batters after the sixth inning, and 18 of the 21 hits against him were singles. Another creepy thing about those longer games for him was the fact that his strikeout rate went UP after the seventh inning. But he rarely pitched that late. But we can forgive him for that, as he was forty years old at the time.

So if a starting pitcher isn't going deep into games, and he is lit up when he does, then he had better have a little luck on his side and a lot of help from his friends going for him. I don't think he had either in 1987.


What follows is support for the above statement.

The Astros scored 648 and allowed 678 runs that year to finish 76-86 which was close to their pythagorean expectation of 78-84.

The offense scored zero or one run in his starts eleven times (three shutouts). That hurts. Thirteen times the support was two, three, or four runs. Ugh...

That team did fairly well in games where the winner scored only a few runs.

If the Astros scored one, two, or three runs, their record was 14-56.

If their opponent scored one, two, or three runs, their opponent's record was 9-57. So their chances were slightly better than the other team's chances.

What if they knew going on that they would score exactly four runs?

The Astros clocked in at 12-11 in those games.

Their opponents were 17-6 in those games. That is a pretty decent spread. Their opponents were sitting pretty if they only scored four runs.

Here is the breakdown if the teams knew going in that they would score either three or four runs:

Astros score three or four runs, their record: 19-24

Opponents score three or four runs, their record: 25-17

Most teams lose when they only score two. Most teams win when they score five. It's those 3 or 4 runs scored games that tell the tale.

What does this have to do with Nolan Ryan's record that year?

Well, I can tell you that Astros were very good as scoring runs in the first, seventh, sixth, and fifth innings that year. Those were their four best innings, in that order. I think the offense was bailing out a lot of pitchers in those sixth and seventh innings (bailing out sometimes meant scoring a run to tie the game at 2-2 or 3-3.)

Going into the sixth inning tied here is what the Astros managed at the end of the sixth inning:

Took the lead: 8 times
Still tied: 8 times
Fell behind: 4 times

Going into the seventh inning tied... here is what the Astros managed at the end of the seventh inning:

Took the lead: 8 times
Still tied: 8 times
Fell behind: 4 times

So, (noting the small sample size) They were twice as likely to take the lead in a tie game in the 6th and 7th innings as they were to fall behind.

But the starting pitcher has to stick around through the seventh inning to reap that benefit in a game which is tied after five innings and he looks to be headed towards a no-decision.

So what was the score when Ryan was pulled?

April: down 3, down 2, down 1, tied
May: ahead nine, down 1, ahead 1, down 1, down 1, down 1
June: down 3, ahead 1, ahead 4, down 3, down 4, down 4
July: down 1, down 1, tied, down 1, down 1
August: tied, ahead 2, ahead 2, ahead 2, ahead 2, down 2
Sept/Oct: ahead 2, ahead 2, ahead 7, tied, ahead 2, down 3, down 2

Number of times he left before pitching five innings, but hit team was ahead anyway... none.

Number of times he left when his team was losing, and he got the win anyway when they scored enough runs in the bottom of the inning... once (opening day against my Dodgers).

So he got one unexpected win. Was he "robbed" of a few expected wins? Did his team fail to bail him out when he departed and he was only down by one run?

In his 34 starts:
Record if there was no scoring after he left: 12-18 with four no decisions.

Number of times he left the game only down by one run: nine

Innings pitched in those games:
4.2
6.0
6.0
6.2
6.0
7.0
7.0
5.0
5.1

He went on to lose seven of those games, as the offense only bailed him out twice. That is lower than expected.

When he left a tie game, the Astros finished 1-3. Ugh, so even staying in the game may not have led to a win.

The Astros were 4-6 as a team in Ryan's no-decisions... so even if he did not depart early, perhaps he still would not have improved his won-lost record. No help there...

The Astros were 12-22 in those games he pitched. That's awful. Take note that the Astros were 64-64 in all other games.

This 1987 season may just be one of those bad luck seasons.

In his no-decisions, he left with the lead five times, the game was tied three times, and he was down one run twice.

So the bullpen hurt hit him five times, while the offense helped him twice. Again, no help there from a team which was .500 in games started by other pitchers. His bullpen hurt him more often than his offense bailed him out.

Close to a best case scenario: Let's say he won every game in which he left with the lead, and got a no decision on any game he left with a tie or was down by one, and lost every game where he left down by more than one run... that would make for an interesting season. Only once did he leave a tie game and he ended up winning (opening day).

So what would his record be by month in that somewhat perfect storm? Situation: The bullpen holds every lead, and the offense bails him out of EVERY one run deficit, and hell.. we'll still give him that opening day win:

Wins-Losses-No Decisions per month as above:
1-1-2
2-0-4
2-0-4
0-0-5
4-1-1
4-2-1

His record would be 13-4, getting the decision in only half of his 34 starts.

That's with great help from his teammates after he leaves the field. But no one has perfect teammates. So it may have needed to pitch one more inning now and then to not only allow the offense one more chance, but also help avoid some blown saves.

Five times he left with the lead and got a no decision. Here is how much he pitched in those games:
7.0
6.1
6.0
6.0
5.1

Even with those five wins, he goes 13-16. He just did not catch any breaks... very unlucky.

Aside from wins and losses, in 34 starts, how many times do you think he gave up more than three runs? Not earned runs, but runs-runs? How many times did he watch the other team plate four while he was out there?

Only six times, in runs: Eight, Five, Four, Four, Four, Four

Twenty-eight times he let in three or fewer runs. Dang..... and he still only "won" eight games.

Number runs scored by his team in his 34 starts:
No runs: three games
One run: eight games (including three of the last four starts)
Two runs: five games
Three runs: four games
Four runs: four games
Five runs: three games
Six or more: seven games

Here is how unlucky he was:

The pitchers for the other team allowed three or fewer runs in a game 20 times.
He allowed three or fewer runs 28 times. (not complete games mind you, so the other team still hit against bullpen after he left).

And he still finished with a 8-16 record. This because (my suggestion) that he did not pitch late enough into games and he was hurt more than he was helped by circumstances in games after he departed.

He went 12-8 in 1986 and 12-11 in 1988. He had a near league average ERA in both seasons, yet would up with a better "record" even with the team scoring fewer runs in 1988, and about the same in 1986 when they were a very good team (30 games over .500)

Go figure...

[leewileyfan]

Coach has delved farther into Ryan's '87 season than I ever would; but in the context of pitcher performance vs. W-L records, here's how Ryan's performance & expectation measure up against his 8-16 record.

1. Nolan Ryan.............................................. .............League

WHIP 1.139............................................. .............WHIP 1.372

If we start there and progress from that/IP base and simply factor in 110 yeras of historic WHIP "units" by a factor of 1.35 "bases" [to accommodate extra base hits], we can get a rough idea of Ryan vs. League, independent of W-L records. This allows us to deal only with expectation.

Then, historically, bases get distilled into "runs scored" @ .270 rate. Brett and I have exchanged posts on how to adjust this final factor to suit individual pitchers. What I use here, on a trial basis, in pitcher NET K>BB/9IP relative to 27 outs.

2. Ryan pitched 211.2 innings. I then take the IP, transferred to League WHIP and Ryan WHIP.

3. In 1987 Ryan had 7.8 NET K's > BB/9IP = 7.8/27 = 28.89%. This is dicounted from the .270 historic scoring rate and for Ryan adjusts to .192.

In 1987, NL K>BB/9IP was 2.6 or 9.63%, which reduces .270 to .244.

4. Applying the numbers:

Ryan = 211.2*1.139*1.35*.192 = 62.35 runs

NL - 211.2*1.372*1.35*.244 = 95.45 runs, suggestin6 that Ryan was +33.1 runs better than League average for the portion of innings he pitched.

5. By my defense metric, HOU was the worst defensive team in the 1987 N.L. @ -39.07 defense runs. This is unadjusted for pitching staff K's vs. League. Adjusted, the defense runs = -20.90.

6. Factor in batting runs vs. League and HOU = -83. Combined, defense + batting = -100 --110 or = -10/-11 wins.

These inner dynamics of Ryan's 1987 season help explain why he was 5th in Cy Young voting, despite an 8-16 record.

[Los Bravos]

I have always wondered about Clemens' Yankee years. Popular viewpoint says that Clemens started using in 1997 or 1998, which if one likes to look at statistical improvement would make sense (he was still great from 1994-1996, but not quite as much as 1986-1993 or 1997-1998), but the worst (relatively, of course) period of his career came as a Yankee after he was supposedly several years into usage. His numbers did not take off into absurd-land again until he went to Houston.

We have some options here:

1. He stopped using when he got to N.Y., then started again in Houston (seems a little problematic)

2. Even on roids, age caught up with him in NY and his performance suffered, but the NL was so much worse than the AL, he dominated despite age (too many ifs - not likely at all)

4. We have no idea how much impact any PED can have with any given player. Clemens had a unique experience with PEDS, and looking at career paths to determine usage is a shaky premise since there are many, many PED and non-PED careers that do not follow expected trends.Those are all plausible. We'll never know unless the fool comes clean in a last ditch effort to salvage his place in history.
Also - looking at Clemen's pitch repertoire, he increased his off-speed pitches (which of course decreased his fastball usage) in Houston, at least. That doesn't seem like the type who refused to lose a power-pitcher image. In fact, his splitter/sinker in Houston became a primary pitch (and one of his most successful.)I consider the split finger a power pitch.
Koufax had one of the league's best offenses behind him. It wasn't until his last year or so that they started to slip from the top. The Dodgers played at Dodger Stadium and it did much to hide the strength of their offense while showcasing their pitching.Fine. It's withdrawn. I'll never say it again.

[thefeckcampaign]

Here are two of my favorite myths, misconceptions about pitchers and win totals.

Pitchers somehow control how many innings they pitch or the number of games they start. Every time a manager goes to the mound to remove his ace with two outs in the seventh, that pitcher has a look on his face like he does not want to leave the game. He tips his hat to the crowd and walks disappointedly to the dugout. We all see this, right? I mean, we aren’t lowly statheads who don’t actually watch the games and spend all our days in our parents’ basements playing WarCraft, right?

Anyway, despite the fact that we all know this people still throw out the old reliable canard of “if Greinke would pitch more innings, he’d have more wins” as if Greinke removed himself from those games. Trust me, Zack Greinke wanted to finish those games. To say that pitcher X should have pitched more innings is a straw man when evaluating the skills of pitcher X because he has very little control over how many innings he pitches in the first place.Sorry to be late to this discussion, but you are quoting me. I only say that in the fact that people are comparing Greinke to guys who did finish, who do have more wins. Whether he wanted to finish or not is not really my deal, it's whether or not these same great stats he has now would be the same or would he have given up more runs later on. In Greinke's case, are the managers actually smarter today? I can't believe I am actually saying that. ;)


My second beef is less of a myth and more of just an internal inconsistency in argumentation. So, if wins imply that a pitcher pitched well for his team to win, why does it matter so much that pitchers today usually pitch fewer innings? If 6 inning wins aren’t a good enough barometer of a pitcher’s success, then that seems to be a flaw with the stat, not the pitcher earning that stat. When people bash these pitchers and their six inning wins, they are bashing the win stat, but then those same people often turn around and extol the virtues of high win totals, even though they have already implicitly acknowledged the arbitrary nature of what it takes to get a win. (Also, why use high win totals as a proxy for durability or ability to go deep into games when we have very straightforward stats established just to measure those things, like, I dunno, innings pitched. And finally, the same people who bash the six-inning, three-run quality start, are the same people to assign positive value of the pitcher who goes eight and gives up five, while earning a 6-5 win.
I think the point is with someone in Greinke's case and the season he had being on a similar if not better team than Carlton's '72 Phillies, he just did not get the total amount of decisions.

[thefeckcampaign]

Here's a question I have, are there any 300 game winners that do not stand out as truly the best in the game as that stat hails them to be? If so, are there 300 GW that with these new stats available can be proven to be worse pitchers than some pitchers with significantly lower win totals? Say someone with 250 wins or lower?

[brett]

Sorry to be late to this discussion, but you are quoting me. I only say that in the fact that people are comparing Greinke to guys who did finish, who do have more wins. Whether he wanted to finish or not is not really my deal, it's whether or not these same great stats he has now would be the same or would he have given up more runs later on. In Greinke's case, are the managers actually smarter today? I can't believe I am actually saying that. ;)


Managers have definitely come to utilize starters and relievers in a way that minimizes total run production. Even a below average reliever is going to prevent runs compared to a good starter the 4th time through the lineup.

[leewileyfan]

Here's a question I have, are there any 300 game winners that do not stand out as truly the best in the game as that stat hails them to be? If so, are there 300 GW that with these new stats available can be proven to be worse pitchers than some pitchers with significantly lower win totals? Say someone with 250 wins or lower?

If you permit a bit more leeway, allowing for comps between HoF pitchers with fewer than 300 wins against guys not in the HoF [or likely to get in], I'd suggest quite a few who, when they were active, were better, more efficient, and consistent than HoF peers.

Some got late starts, suffered injury, had careers interrupted by WW; and some went on and on with losing ballclubs:

Lon Warneke
Ted Wilks
Bucky Walters
Paul Derringer
Spud Chandler
Dutch Leonard
Smoky Joe Wood
Tex Hughson
Mort Cooper
Sal Maglie
Max Lanier
Jimmy Key
Ron Guidry
Andy Messersmith
Harry Brecheen
Kevin Appier
Don Newcombe

.... on the relief side, we don't see too many rave reviews for Tom Henke, either.

[brett]

Saberhagen
Hershizer
Stieb
Cone

[BiZmaRK]

Greinke's record when the Royals scored zero runs was 0-3 in 4 starts.
Greinke's record when the Royals scored one run was 0-1 in 2 starts.
Greinke's record when the Royals scored two runs was 3-1 in 6 starts.
Greinke's record when the Royals scored three runs was 3-1 in 7 starts.
Greinke's record when the Royals scored 4+ runs was 10-2 in 14 starts.

Carlton's record when the Phillies scored zero runs was 0-2 in 3 starts.
Carlton's record when the Phillies scored one run was 2-3 in 5 starts.
Carlton's record when the Phillies scored two runs was 7-2 in 9 starts.
Carlton's record when the Phillies scored three runs was 4-2 in 6 starts.
Carlton's record when the Phillies scored 4+ runs was 14-1 in 18 starts.

Carlton started eight more games than Greinke, and nine more than Greinke in which his team scored a run. So that's part of it. But if Greinke had won games at the same rate Carlton did given each level of offensive support, he'd have won 21 games instead of 16. That's one way to look at it.

Another way to look at it is that the bullpen blew three games where Greinke left with the lead.
An easy fix to the problem would be instead of using actual runs scored by the pitcher's team, use the average number of runs scored league wide over the number of innings he pitches. That way you can appropriately compare Greinke to Sabathia. Additionally, do away with the requirement that a pitcher must go 5 innings to get a win. He doesn't have to go 5 innings to get a loss, so why the need to go 5 for a win?

Let's say you start the game, go 4 innings and leave with your team up 11-0. Each reliever after you goes 1 inning and gives up 2 runs & your team wins 11-10. You can't get the win - even though you were clearly the best pitcher for your team that day.

Pitchers W-L in baseball is so outrageously useless that it even trumps the GWG in hockey when it comes to idiotic stats.

[STLCards2]

Pitchers W-L in baseball is so outrageously useless that it even trumps the GWG in hockey when it comes to idiotic stats.

Well, there is a .65 correlation between W% and ERA (a much higher correlation than run support and W%), so to say it is "outrageously useless" is not fair or accurate.

Being literal, for pitchers' W-L records to be "outrageoulsy useless", then a 18-10 pitcher would have an equal likelihood of having a 3.00 ERA as a 10-18 pitcher, with obviously isn't the case.

What's true is: that there are many things a lot better at determining pitcher effectiveness than W-L record, so why ever it.

[BiZmaRK]

Well, there is a .65 correlation between W% and ERA (a much higher correlation than run support and W%), so to say it is "outrageously useless" is not fair or accurate.

Being literal, for pitchers' W-L records to be "outrageoulsy useless", then a 18-10 pitcher would have an equal likelihood of having a 3.00 ERA as a 10-18 pitcher, with obviously isn't the case.

What's true is: that there are many things a lot better at determining pitcher effectiveness than W-L record, so why ever it.
It's "outrageously useless" because there are so many inherent flaws in the stat as well as there being many other more effective methods of measuring the same thing that they're attempting to measure using pitchers W-L.

[leewileyfan]

What's true is: that there are many things a lot better at determining pitcher effectiveness than W-L record, so why ever it.

I have seen the term "many things" [and/or the like] attributed to unnamed statistical data presumed superior to W-L. However, in these recent exchanges, I have yet to see one of these superior data sets specifically cited, defined or exemplified to show such superiority.

I am basically a traditionalist, slow to accept drastic changes is game rules and dynamics, especially when they incline to discount former valued skills and exaggerate skill sets assumed to be more precious than the seem to be in actual practice.

When I was a kid, I wondered why no managers had ever considered the limted start concept: three innings per pitcher per game, with > 3IP only of pitcher was outstanding through three. I believed that such an approach would allow pitchers to intensify focus, maximize effort within a psychological comfort zone; and maybe cash in on a "use it or lose it" strategy in which two-three days rest would be a maximum.

A top gun might then make upward of 50 starts, averaging, say 5 IP/start and piling up a max of 250 IP in a season.
Of course the immediate question in my mind was: What about the 5 IP required for a win.

Bottom line for me then was: Change the rules.

Now that the game is shifting to low inning starts, I am somehow resistant to a change that approximates my [I]off-the-wall thinking of over 65 years ago. I realize that's just tradition. The simple dropping of an adjective, like "starting," "set-up" or "closing" might be the most practical solution, Call them pitchers. Then give them a rating based on the before-after states of the 24 base-out situations in which they entered and departed each game entered.

Over a season, each pitcher would have a net number for the season, + or -, which would record statistical expectation against actual performance.

The net of the entire staff would then be the team's pitching contribution. Defense and batting contributions could them be added to the mix.

Done correctly, the aggregate should resemble W-L, which would be a team stat only at season's end.

[STLCards2]

I have seen the term "many things" [and/or the like] attributed to unnamed statistical data presumed superior to W-L. However, in these recent exchanges, I have yet to see one of these superior data sets specifically cited, defined or exemplified to show such superiority.

.

I doubt you would push WHIP and "legacy points" or whatever it is, if you didn't think they were fundamentaly better than W-L record in isolation. Why not just use W-L record if you don't think there is better out there? The fact that you used a "WHIP modified by a K-based factor" to change W-L records earlier in this thread proves that there are thigs you think work better than unadjusted W-L record too.

"These inner dynamics of Ryan's 1987 season help explain why he was 5th in Cy Young voting, despite an 8-16 record."

Why use any "inner dynamics" if you feel that W-L record holds up well without them?

[leewileyfan]

I doubt you would push WHIP and "legacy points" or whatever it is, if you didn't think they were fundamentaly better than W-L record in isolation. Why not just use W-L record if you don't think there is better out there? The fact that you used a "WHIP modified by a K-based factor" to change W-L records earlier in this thread proves that there are thigs you think work better than unadjusted W-L record too.

First, I made reference to many citations of "better" or "superior" stats to W-L records, with no specifics as examples.

Is your response above responsive to my observations of what others have alleged? I have basically supported W-L as a traditional and useful starting point. That is, until my last post on the topic in which I confessed a fascination with 3 pitchers per game as a strategy never tried [to my knowledge].

Your response seems to imply that, promoting WHIP as an essential revealer of pitcher effectiveness, disqualifies my from seeing W-L as useful.

"These inner dynamics of Ryan's 1987 season help explain why he was 5th in Cy Young voting, despite an 8-16 record."


Why use any "inner dynamics" if you feel that W-L record holds up well without them?

As I stated above, you see further exploration as a disqualifier for previous values accepted. I can accept A as a decent basic reference, and having some value; but, if I see weaknesses and explore those weaknesses, I do not throw the baby out with the bathwater.

Ryan's 8-16 bears scrutiny; and WHIP helps meliorate the drubbing Ryan takes on W-L alone, when other dynamics beyond his immediate control worsened his record.

I also provided defense runs and batting runs which are, at best, secondary and tertiary to pitcher evaluation.

[curveball]

Well, there is a .65 correlation between W% and ERA (a much higher correlation than run support and W%), so to say it is "outrageously useless" is not fair or accurate.

Being literal, for pitchers' W-L records to be "outrageoulsy useless", then a 18-10 pitcher would have an equal likelihood of having a 3.00 ERA as a 10-18 pitcher, with obviously isn't the case.

What's true is: that there are many things a lot better at determining pitcher effectiveness than W-L record, so why ever it.


Just curious, but is the correlation between W% and era really that much higher than run support and W%?

0 run support obviously means 0 win %.
1 run support probably means a 3% win percentage.
If a pitcher gets 4 runs of support, I am pretty sure even the average pitcher would win at least 50% of his games.
At 5 runs of support, I think the average pitcher would win close to 70% of his games.

I would have to agree with BizmarK that W/L % is almost utterly useless. There doesn't even have to be any correlation really with W% and era.

What would happen if you put the greatest pitchers ever (W Johnson, Clemens, Maddux, etc...) and put them on a team of baseball players that played decent defense, but only had a cumulative batting average of .050. This team may never score a run, so Johnson, Clemens, and Maddux could conceivably go 0 and 40, but have great eras because that at least would remain constant.

That is the most important thing. Good pitching will always remain constant. Maddux will always be a great pitcher regardless of what team he plays for, as long as they play adequate defense of course. But if Maddux plays for a team that usually scores him no runs, then he will never be a winning pitcher.

Pitcher's really have no control over the defenses that play behind them or the runs their team scores for them. That is why I really believe that W/L are really virtually useless.

The only reason there is a .65 correlation as you state is because teams usually hover around the average run support, give or take a run. But if that wasn't the norm, and you had some really weak hitting teams that scored virtually no runs, then that correlation would take a severe hit.

[curveball]

An easy fix to the problem would be instead of using actual runs scored by the pitcher's team, use the average number of runs scored league wide over the number of innings he pitches. That way you can appropriately compare Greinke to Sabathia. Additionally, do away with the requirement that a pitcher must go 5 innings to get a win. He doesn't have to go 5 innings to get a loss, so why the need to go 5 for a win?

Let's say you start the game, go 4 innings and leave with your team up 11-0. Each reliever after you goes 1 inning and gives up 2 runs & your team wins 11-10. You can't get the win - even though you were clearly the best pitcher for your team that day.

Pitchers W-L in baseball is so outrageously useless that it even trumps the GWG in hockey when it comes to idiotic stats.


Agreed.

You can conceivably have a pitcher who has never lost, so let's say his record is 1000 and 0, but he could be a horrible pitcher.

You can have a pitcher who has never won, so let's say his record is 0 and 1000, but he could be a million times better than the undefeated pitcher.

That should sum up how useless W/L records can be.

[brett]

Just curious, but is the correlation between W% and era really that much higher than run support and W%?

0 run support obviously means 0 win %.
1 run support probably means a 3% win percentage.
If a pitcher gets 4 runs of support, I am pretty sure even the average pitcher would win at least 50% of his games.
At 5 runs of support, I think the average pitcher would win close to 70% of his games.

I would have to agree with BizmarK that W/L % is almost utterly useless. There doesn't even have to be any correlation really with W% and era.

What would happen if you put the greatest pitchers ever (W Johnson, Clemens, Maddux, etc...) and put them on a team of baseball players that played decent defense, but only had a cumulative batting average of .050. This team may never score a run, so Johnson, Clemens, and Maddux could conceivably go 0 and 40, but have great eras because that at least would remain constant.

That is the most important thing. Good pitching will always remain constant. Maddux will always be a great pitcher regardless of what team he plays for, as long as they play adequate defense of course. But if Maddux plays for a team that usually scores him no runs, then he will never be a winning pitcher.

Pitcher's really have no control over the defenses that play behind them or the runs their team scores for them. That is why I really believe that W/L are really virtually useless.

The only reason there is a .65 correlation as you state is because teams usually hover around the average run support, give or take a run. But if that wasn't the norm, and you had some really weak hitting teams that scored virtually no runs, then that correlation would take a severe hit.

I think that a .65 correlation means that about 40% of winning percentage can be explained by raw ERA. Not great but meaningful.

[STLCards2]

Just curious, but is the correlation between W% and era really that much higher than run support and W%?


.

I think it is .5 - relatively close, but still not as good.

To say that W-L is completely useless would imply that given equal defense and offensive support, that W-L record would be random, which it certainly would not be. Like brett said - not great (or even good) but meaningful.

Clearly, better pitchers have better W-L records that their poor pitcher counterparts on high- offensive teams, middle offensive teams, and low offensive teams. If this weren't the case, than you would see near total randomness of W-L records for starting pitchers who share the same teamates.

Of course Greg Maddux's record could be worse than Jose Lima's if Maddux was on the worst team ever and Lima on the best team ever. But to say that W-L is "completely useless" would imply that their record would predictably be the same if they were both on the "best team ever". If both played on the 1927 Yankees and Maddux was 27-6 and Lima 18-15, would you not say that their W-L records had anything to do with their ability? Of if Maddux was 12-16 on the 2003 Tigers and Lima 6-20?

[leewileyfan]

You can conceivably have a pitcher who has never lost, so let's say his record is 1000 and 0, but he could be a horrible pitcher.

Really? This statement avoids taking on the relative value W-L as a valuable starting statistic by throwing a virtually impossible supposition [and example] into the mix.


You can have a pitcher who has never won, so let's say his record is 0 and 1000, but he could be a million times better than the undefeated pitcher.

This takes the improbable [to say the very least] and raises it to the nth power, presumably to emphasize a point not made.

Any fan of MLB, interested and even modestly motivated might see Kaiser Wilhelm's [nope, didn't make it up] losing record for the 1908 BRO team and not draw a conclusion on his pitching without looking up some form of context first.

Seeing BRO had a team BA of .213 that year might bring some empathy toward a guy with a 12-17 record and an ERA+ >125.

Tossing on virtually impossible situations and exclaiming their virtues to denigrate W-L is hardly a convincing way to make a salient point.

Stating further shades of relevance factors of other stats in isolation vs. W-L in isolation also does little to resolve debate over W-L.

W-L records, at the bottom line, are one extension of pitching data. Pitching data are yet another subset of team defense, in which the efforts of eight other players are subsumed and folded into a supporting role for the one guy in the middle of the diamond. Even those two components still comprise only half the W-L picture, the other half being the part when the pitcher's team gets its at-bats.

Burdening W-L records for pitchers with the weight of all that is simply to illustrate that the evaluator has failed to realize the function of the record in the larger context.


That should sum up how useless W/L records can be.

Not a bit.

[curveball]

I think it is .5 - relatively close, but still not as good.

To say that W-L is completely useless would imply that given equal defense and offensive support, that W-L record would be random, which it certainly would not be. Like brett said - not great (or even good) but meaningful.

Clearly, better pitchers have better W-L records that their poor pitcher counterparts on high- offensive teams, middle offensive teams, and low offensive teams. If this weren't the case, than you would see near total randomness of W-L records for starting pitchers who share the same teamates.

Of course Greg Maddux's record could be worse than Jose Lima's if Maddux was on the worst team ever and Lima on the best team ever. But to say that W-L is "completely useless" would imply that their record would predictably be the same if they were both on the "best team ever". If both played on the 1927 Yankees and Maddux was 27-6 and Lima 18-15, would you not say that their W-L records had anything to do with their ability? Of if Maddux was 12-16 on the 2003 Tigers and Lima 6-20?


If both played on the 1927 Yankees and Maddux was 27-6 and Lima 18-15, their W/L records would mean absolutely nothing to me until I looked at their
eras and run support received. Knowing what I already know about Maddux, and Lima, I would expect Maddux to have the better W/L record because I expect him to have the lower era because he is the better pitcher. I would also assume that they both should receive similar run support.

Without knowing the abilities of Maddux and Lima, I could already come to conclusions about them based on their W/L records. Maddux definitely has either,

a) lower era (of course, I assume defenses were similar)
b) better run support

It would be impossible for Maddux to have both,

a) higher era
b) lower run support


What if Maddux was 0-27, and Lima 18-15? Would you bet your life that Lima was the better pitcher?

Just looking at these W/L records, I an certain that,

a) Lima received better run support

but I am not certain that Lima has,

a) the lower era (assuming similar defenses again)

Of course, this situation is highly unlikely in MLB, but it simply illustrates the point that a great pitcher cannot win any games if a team does not or chooses not to score any runs for him. This great pitcher will still be a great pitcher because that will always remain constant regardless of how many games he wins. The greatest variable is his run support. That is why I never look at W/L records when judging pitchers.

If there was a 100 piece jigsaw puzzle that unravels how good a pitcher is,
then W/L record would be the last piece of the puzzle, or wouldn't be needed at all because the 99 other pieces painted a much clearer picture.

It has happened many times in baseball history that two pitchers on the same team have had W/L records that do not correspond to how well they pitched. Pitcher A could have had an era that was a whole run lower than pitcher B, yet pitcher B had a much more impressive W/L record, and that was all because of run support.

There are many instances of pitchers with sparkling yearly eras, yet they had losing records or unflattering winning percentages. How many pitchers, no matter how lousy they were, had losing records if they received an average 5 or more runs of support per start?

[curveball]

Really? This statement avoids taking on the relative value W-L as a valuable starting statistic by throwing a virtually impossible supposition [and example] into the mix.



This takes the improbable [to say the very least] and raises it to the nth power, presumably to emphasize a point not made.

Any fan of MLB, interested and even modestly motivated might see Kaiser Wilhelm's [nope, didn't make it up] losing record for the 1908 BRO team and not draw a conclusion on his pitching without looking up some form of context first.

Seeing BRO had a team BA of .213 that year might bring some empathy toward a guy with a 12-17 record and an ERA+ >125.

Tossing on virtually impossible situations and exclaiming their virtues to denigrate W-L is hardly a convincing way to make a salient point.

Stating further shades of relevance factors of other stats in isolation vs. W-L in isolation also does little to resolve debate over W-L.

W-L records, at the bottom line, are one extension of pitching data. Pitching data are yet another subset of team defense, in which the efforts of eight other players are subsumed and folded into a supporting role for the one guy in the middle of the diamond. Even those two components still comprise only half the W-L picture, the other half being the part when the pitcher's team gets its at-bats.

Burdening W-L records for pitchers with the weight of all that is simply to illustrate that the evaluator has failed to realize the function of the record in the larger context.



Not a bit.


I really don't know how to make this any simpler.

A pitcher's job is to prevent runs from scoring. If he does this well, then he has done his job.

A pitcher is absolutely helpless as to how good his defense is, or how good his team's offense is, as far as scoring runs for him, and rewarding him with a win goes.

A pitcher can only do so much, and era (adjusted for defense, ballpark, etc...) would be the best and only way to judge his performances. Wins and losses don't really matter at all.


Take Usain Bolt as an example. He is the world's fastest man. He relies only on himself to win races, and he does win his races with relative ease. He is the "best" and "fastest" sprinter in the world period.

Let us now surround Bolt with some teammates, as there are other races, like the 4 x 100 relay. Let us argue that Jamaica doesn't have any other fast sprinters, and Bolt has never won a 4 x 100 relay race for his native country. So bolt is 0 and 50 in relay races. Does this now make him a lousy or slow runner because he has never won a 4 x 100 relay race in his career? In fact Bolt and his teammates have never even made a relay final in his career because his teammates were so slow. This does not take away from the fact that Bolt is still the world's best and fastest sprinter.

Similarly, take a pitcher like Greg Maddux. By himself, he is a great pitcher. That will always remain constant. The only problem is that Maddux cannot win baseball games by himself, like Bolt can, in a 100 m race.

Maddux has to surround himself with teammates, who can hopefully play solid defense behind him, and hitters that can score him plenty of runs. No matter how good Maddux is, he needs help to win baseball games, and so does Bolt, when he decides to run in relay races, as he needs 3 other teammates.

In baseball, wins are a team stat period, yet some guy arbitrarily decided to reward pitchers with wins and losses, and because of this, too many people still believe that they are actually important.

Maddux is a great pitcher. Bolt is the fastest runner. Force them to have teammates, and they may no longer be winners, but it isn't because Maddux has all of a sudden become less great, or Bolt has become slower.

[digglahhh]

Again, this notion of the win being "worth" is now muddled in weeds of semantics. So, let's stick to the "far superior" data sets idea. If you were a General Manager, and last year when your scouts sent you the information availabled on the pitchers available for acquisition, you were given the following stats:

IP:
K/9
BB/9
HR/9
Opp BA/OBP/SLG
WHIP
ERA+
BABIP
Win Shares
FIP
FB/GB/LD%
Following year Marcels
W-L

For some reason, you must sacrifice one of these stats on next year's report. Would you choose something other than W-L?

[curveball]

Again, this notion of the win being "worth" is now muddled in weeds of semantics. So, let's stick to the "far superior" data sets idea. If you were a General Manager, and last year when your scouts sent you the information availabled on the pitchers available for acquisition, you were given the following stats:

IP:
K/9
BB/9
HR/9
Opp BA/OBP/SLG
WHIP
ERA+
BABIP
Win Shares
FIP
FB/GB/LD%
Following year Marcels
W-L

For some reason, you must sacrifice one of these stats on next year's report. Would you choose something other than W-L?


I think it would be easier to compile a list of stats that you think W/L is more valuable than. That would be a much shorter list, that is if you could even come up with anything. :)

[leewileyfan]

I really don't know how to make this any simpler.


A pitcher's job is to prevent runs from scoring. If he does this well, then he has done his job.

Red flag #1 in any discussion or debate: I really don't know how to make this any simpler is a dead giveaway to a weak argument coming from someone who believes condescension to a slower [dimmer] audience will insure a convincing start. NOT.


A pitcher is absolutely helpless as to how good his defense is, or how good his team's offense is, as far as scoring runs for him, and rewarding him with a win goes.

This is absolutely indisputable. The pitcher is not the team's GM or its owner. However, we are talking MLB; and even the worst team imaginable will win 25% of its games. In a 154 game schedule, that projects 38-116. In a 162 game schedule, it's 40-122. [Pretty bad].

However, your point is focused not on the pitcher but on his lack of defensive support and batting support. We are discussing a stat as it particularly relates to pitchers, W-L.

What, in fact, is a team's record in any season? It's W-L, as it should be. However, is anybody on this board suggesting the uselessness of the statistic in this team concept.? No, we are mis-applying the stat, first conceived as a promotional drawing card for starting pitchers [home & visitors], as if it were the sole evaluator for pitcher effectiveness. No informed fan has ever believed this to be the case. It's a subset that, when added for all individual pitchers, exactly equals the team's record.


A pitcher can only do so much, and era (adjusted for defense, ballpark, etc...) would be the best and only way to judge his performances. Wins and losses don't really matter at all.

This is a ALL - or NOTHING at all argument that holds no water in the proper context of W-L function. One can, for example, select all 30 win seasons since 1901 as a starting point for a study attempting to evaluate thos performances against eachother. All W-L does in to provide a common denomonator for "qualifiers." Another study may throw out the 30 wins prerequisite and substitute 275 IP as the qualifying basis. W-L then disappears from study #2, unless it is selected as a marker for each of the best seasons.


Take Usain Bolt as an example. He is the world's fastest man. He relies only on himself to win races, and he does win his races with relative ease. He is the "best" and "fastest" sprinter in the world period.

OK. But why must we shift to individual sprints as a salient debating point, when we are talking about a team sport that dilutes the individual into the larger contexts of team, position, role?


Let us now surround Bolt with some teammates, as there are other races, like the 4 x 100 relay. Let us argue that Jamaica doesn't have any other fast sprinters, and Bolt has never won a 4 x 100 relay race for his native country.

Your argument goes off the rails on two points: Bolt is not weighted down with sluggish 'mates in a team effort; Jamaica is not noted for its marked lack of swift sprinters.


So bolt is 0 and 50 in relay races. Does this now make him a lousy or slow runner because he has never won a 4 x 100 relay race in his career? In fact Bolt and his teammates have never even made a relay final in his career because his teammates were so slow.

Why are your examples always devoted to such extremes of ineptitude? Exaggerations do not make illogical arguing points any more convincing.


This does not take away from the fact that Bolt is still the world's best and fastest sprinter.

For that, I'll take your word as gospel. I am not a big follower of track & field.


Similarly, take a pitcher like Greg Maddux. By himself, he is a great pitcher. That will always remain constant. The only problem is that Maddux cannot win baseball games by himself, like Bolt can, in a 100 m race.

At 300+ wins, wins do take on a smidgeon of significance.


Maddux has to surround himself with teammates, who can hopefully play solid defense behind him, and hitters that can score him plenty of runs. No matter how good Maddux is, he needs help to win baseball games, and so does Bolt, when he decides to run in relay races, as he needs 3 other teammates.

Maddux does not surround himself with teammates. The front office surrounds him with teammates they have selected. At that point, any comparison of Maddux and Bolt is irrational and invalid.


In baseball, wins are a team stat period, yet some guy arbitrarily decided to reward pitchers with wins and losses, and because of this, too many people still believe that they are actually important.

I'd be it wasn't some single guy, but rather some combination of scorekeepers, sports writers, team statisticians and media ad men who came up with a concept that comported team wins exactly with the sum of individual pitching staff wins.

What you are staunchly condemning is not W-L on its own merits, but the distribution of credits [sometimes awarded to less-deserving performers within particualar games. You are tossing the baby out with the bathwater.


Maddux is a great pitcher. Bolt is the fastest runner.

Agreed.

[leewileyfan]

Again, this notion of the win being "worth" is now muddled in weeds of semantics. So, let's stick to the "far superior" data sets idea.

What do you have against semantics, which you seem to consider an epithet?


Merriam-Webster:

Main Entry: se·man·tics
Pronunciation: \si-ˈman-tiks\
Function: noun plural but singular or plural in construction
Date: 1893

1 : the study of meanings: a : the historical and psychological study and the classification of changes in the signification of words or forms viewed as factors in linguistic development b (1) : semiotic (2) : a branch of semiotic dealing with the relations between signs and what they refer to and including theories of denotation, extension, naming, and truth


If you were a General Manager, and last year when your scouts sent you the information availabled on the pitchers available for acquisition, you were given the following stats:

IP:
K/9
BB/9
HR/9
Opp BA/OBP/SLG
WHIP
ERA+
BABIP
Win Shares
FIP
FB/GB/LD%
Following year Marcels
W-L

For some reason, you must sacrifice one of these stats on next year's report. Would you choose something other than W-L?

Yep. Let me start with four, just for openers:

1. Marcels
2. Win Shares
3. FB/GB/LD%
4. BABIP

[curveball]

Leewileyfan,

We'll just agree to disagree.

Wins and losses for me will always be a garnish, something that is added for embellishment or decoration, but really having very little value.

There are so many flaws to wins and losses when you try to apply them to individuals, when it is absolutely a team concept.

I am not trying to be condescending. Maybe I am wrong, and someone will someday convince me of that, but to me it is simple that Maddux will always be a great pitcher, and Bolt will always be a fast runner. That fact will always remain constant. If you surround them with teammates, no matter what their respective won/loss records may become, they will still be as great. If this is agreed upon, that both are still great, then what really was the impact of w/l record? Nothing at all.

[STLCards2]

Knowing what I already know about Maddux, and Lima, I would expect Maddux to have the better W/L record because I expect him to have the lower era because he is the better pitcher. I would also assume that they both should receive similar run support.

?

So given similar run support, you are admitting that the better pitcher (Maddux) is more likely to have a good W-L record than the bad pitcher (Lima) because they give up fewer runs. If what you are saying is true, then there is at least some analytical value in W-L, rendering it not "completely" or "outrageously" useless. That is what I have been saying the whole time. Listen, nobody is saying it is a great tool or anything.

As far as digglahh's post, I would trust W-L record for a long career guy more than some of those stats. If a player has a great winning% despite a mediocre FIP, unless he is on a great offensive team every year, I would assume that he is doing something each year that "beats" FIP (GBDP, controling running game, BABIP reduction, low WP, low HB, LOB% reduction, fewer XBH etc.). If were only looking at the previous season or two, I would look at FIP first.

Assuming the aquisition was for a young unknown, If I had to get rid of one of those listed, it would be Win Shares. Second would be Marcels since I could estimate predicted future perfromance without it. BABIP would probably be next, since it is so unstable, there is no way to determine if his previous year's BABIP was "true him" or not. W-L would probably be the next to go. Even though, if I know he was on a horrible team, but his W-L record was very good, I could use it to estimae that he may be better than his W-L record indicates.

[curveball]

So given similar run support, you are admitting that the better pitcher (Maddux) is more likely to have a good W-L record than the bad pitcher (Lima) because they give up fewer runs. If what you are saying is true, then there is at least some analytical value in W-L, rendering it not "completely" or "outrageously" useless. That is what I have been saying the whole time. Listen, nobody is saying it is a great tool or anything.

As far as digglahh's post, I would trust W-L record for a long career guy more than some of those stats. If a player has a great winning% despite a mediocre FIP, unless he is on a great offensive team every year, I would assume that he is doing something each year that "beats" FIP (GBDP, controling running game, BABIP reduction, low WP, low HB, LOB% reduction, fewer XBH etc.). If were only looking at the previous season or two, I would look at FIP first.

Assuming the aquisition was for a young unknown, If I had to get rid of one of those listed, it would be Win Shares. Second would be Marcels since I could estimate predicted future perfromance without it. BABIP would probably be next, since it is so unstable, there is no way to determine if his previous year's BABIP was "true him" or not. W-L would probably be the next to go. Even though, if I know he was on a horrible team, but his W-L record was very good, I could use it to estimae that he may be better than his W-L record indicates.


Just curious, how is W-L more informative than win shares, or any other metric that at least tries to come up with a number stating how valuable a pitcher's season was?

Take Cleveland in 2005. Cliff Lee was 18-5, and Kevin Millwood was 9-11. Is this really a no-brainer because W/L record is so meaningful? I would guess that Millwood would beat Lee in win shares, warp, praa, prar, rsaa, and any other metric that tries to put a value on a pitcher's season. If this is indeed the case, then doesn't that show exactly how useless and misleading W/L record can be, and how those other metrics are still way better than W/L record?

[BiZmaRK]

I think it is .5 - relatively close, but still not as good.

To say that W-L is completely useless would imply that given equal defense and offensive support, that W-L record would be random, which it certainly would not be. Like brett said - not great (or even good) but meaningful.

Clearly, better pitchers have better W-L records that their poor pitcher counterparts on high- offensive teams, middle offensive teams, and low offensive teams. If this weren't the case, than you would see near total randomness of W-L records for starting pitchers who share the same teamates.

Of course Greg Maddux's record could be worse than Jose Lima's if Maddux was on the worst team ever and Lima on the best team ever. But to say that W-L is "completely useless" would imply that their record would predictably be the same if they were both on the "best team ever". If both played on the 1927 Yankees and Maddux was 27-6 and Lima 18-15, would you not say that their W-L records had anything to do with their ability? Of if Maddux was 12-16 on the 2003 Tigers and Lima 6-20?
It makes more sense to simply ask how much control do pitchers have over their win/loss record rather than to analyze any correlation(s). Since they don't have any control over how many runs the team scores and since they don't make out the lineup card or play general manager, all they can control is how much they give up. To say pitchers have control over their W-L record is like saying the Yankees have control over how well the Padres do.

[BiZmaRK]

Tossing on virtually impossible situations and exclaiming their virtues to denigrate W-L is hardly a convincing way to make a salient point.
Please explain how the pitchers W-L record has validity if the pitcher has no control over how many runs his team scores.

W-L records, at the bottom line, are one extension of pitching data. Pitching data are yet another subset of team defense, in which the efforts of eight other players are subsumed and folded into a supporting role for the one guy in the middle of the diamond. Even those two components still comprise only half the W-L picture, the other half being the part when the pitcher's team gets its at-bats.
Then why is W-L record considered an individual stat and not a team stat?

[BiZmaRK]

I really don't know how to make this any simpler.

A pitcher's job is to prevent runs from scoring. If he does this well, then he has done his job.

A pitcher is absolutely helpless as to how good his defense is, or how good his team's offense is, as far as scoring runs for him, and rewarding him with a win goes.

A pitcher can only do so much, and era (adjusted for defense, ballpark, etc...) would be the best and only way to judge his performances. Wins and losses don't really matter at all.


Take Usain Bolt as an example. He is the world's fastest man. He relies only on himself to win races, and he does win his races with relative ease. He is the "best" and "fastest" sprinter in the world period.

Let us now surround Bolt with some teammates, as there are other races, like the 4 x 100 relay. Let us argue that Jamaica doesn't have any other fast sprinters, and Bolt has never won a 4 x 100 relay race for his native country. So bolt is 0 and 50 in relay races. Does this now make him a lousy or slow runner because he has never won a 4 x 100 relay race in his career? In fact Bolt and his teammates have never even made a relay final in his career because his teammates were so slow. This does not take away from the fact that Bolt is still the world's best and fastest sprinter.

Similarly, take a pitcher like Greg Maddux. By himself, he is a great pitcher. That will always remain constant. The only problem is that Maddux cannot win baseball games by himself, like Bolt can, in a 100 m race.

Maddux has to surround himself with teammates, who can hopefully play solid defense behind him, and hitters that can score him plenty of runs. No matter how good Maddux is, he needs help to win baseball games, and so does Bolt, when he decides to run in relay races, as he needs 3 other teammates.

In baseball, wins are a team stat period, yet some guy arbitrarily decided to reward pitchers with wins and losses, and because of this, too many people still believe that they are actually important.

Maddux is a great pitcher. Bolt is the fastest runner. Force them to have teammates, and they may no longer be winners, but it isn't because Maddux has all of a sudden become less great, or Bolt has become slower.
curveball, it's a marketing gimmick designed to reel in the less than informed - or the naive. Not that much different from the marketing gimmicks that get thousands upon thousands of people out of their beds at 2:00 AM on the day after Thanksgiving with the hope they'll be one of the ten people to get a $500 laptop computer for $200. Or the ones which say you should spend 2 to 3 months salary on an engagement ring. Marketing gimmicks designed by those who stand to benefit financially.

[BiZmaRK]

What, in fact, is a team's record in any season? It's W-L, as it should be. However, is anybody on this board suggesting the uselessness of the statistic in this team concept.? No, we are mis-applying the stat, first conceived as a promotional drawing card for starting pitchers [home & visitors], as if it were the sole evaluator for pitcher effectiveness. No informed fan has ever believed this to be the case. It's a subset that, when added for all individual pitchers, exactly equals the team's record.
You seem to be forgetting that W-L is marketed as an individual stat and not a team stat.

[BiZmaRK]

Again, this notion of the win being "worth" is now muddled in weeds of semantics. So, let's stick to the "far superior" data sets idea. If you were a General Manager, and last year when your scouts sent you the information availabled on the pitchers available for acquisition, you were given the following stats:

IP:
K/9
BB/9
HR/9
Opp BA/OBP/SLG
WHIP
ERA+
BABIP
Win Shares
FIP
FB/GB/LD%
Following year Marcels
W-L

For some reason, you must sacrifice one of these stats on next year's report. Would you choose something other than W-L?
If I were a GM, I would look at the team's record in games in which that pitcher started before I'd look at his individual W-L record. Would you or leewileyfan do otherwise?

[leewileyfan]

You seem to be forgetting that W-L is marketed as an individual stat and not a team stat.

First, that's how it started out: draw fans to the ballpark to catch our guy vs. their guy.

Second, the sum of pitcher wins and losses on any given staff = the team's wins and losses - exactly.

Third, subjective interpretation of the stat by individuals [either more or less knowledgeable] should not define the purpose or value of the stat itself.

Marketing is a strategy for selling stuff, often misleading and not too infrequently vulnerable to closer scrutiny. The way a stat is marketed does not define the origins or the attractions of the stat itself.

Like marketing, W-L stands up only to a limited degree of scrutiny. The more one digs, the more one can find nuances, modifications, or biases that inflence compilation of the simply-stated stat itself.

That should not disqualify the statistic itself.

It's a little like the ancient joke, old when I was in maybe the first or second grade, about a haughty, spoiled, and cynical woman who walks into a butcher shop, all filled with doubt and suspicion ......

Lady: Are these chickens fresh?

Butcher: Yes, indeed.

Lady: [Picking up, squeezing, sniffing, and analyzing the carcass from every possible angle and orifice] mutters a doutful "Hmmmmmmmmmmm" and looks cynically at the butcher.

Butcher: Lady. Could [B]you pass that kind of inspection?

[BiZmaRK]

Second, the sum of pitcher wins and losses on any given staff = the team's wins and losses - exactly.
How does this have any bearing on the fact that the pitcher has no control over how many runs his team scores?

You could assign the win or loss to the player who is at bat when the go-ahead run scores and you'd still come up with the sum of players wins and losses equaling the teams wins and losses.


Third, subjective interpretation of the stat by individuals [either more or less knowledgeable] should not define the purpose or value of the stat itself.
It's not subjective that the pitcher doesn't control how many runs his team scores or whether or not he fills out the lineup card.


Marketing is a strategy for selling stuff, often misleading and not too infrequently vulnerable to closer scrutiny. The way a stat is marketed does not define the origins or the attractions of the stat itself.
Maybe not. But the way it is marketed can still put the stat at the forefront of what the news media tries to get us to pay attention to.

[leewileyfan]

How does this have any bearing on the fact that the pitcher has no control over how many runs his team scores?

You could assign the win or loss to the player who is at bat when the go-ahead run scores and you'd still come up with the sum of players wins and losses equaling the teams wins and losses.

The above, in response to my observation that the sum of pitching staff wins [while with a team in a given season] = team wins exactly was just what I said it was: a tidy and convenient congruence between the the pitcher-promotional W-L highlight and the team's W-L record. I made no embellishments.



It's not subjective that the pitcher doesn't control how many runs his team scores or whether or not he fills out the lineup card.

Nor is it subjective when he loses 1-0 in 15 innings. Two pitching workhorses labored against eachother on a good day. One simply had to lose.
Pitcher + Team].


But the way it is marketed can still put the stat at the forefront of what the news media tries to get us to pay attention to.

You are living proof that not all of us are so easily sucked in by media hype. Proudly, I am in that same class of "holdouts." However, I do not find it necessary to downgrade W-L records, if only because - warts & all - it's an attention grabber and it's of long standing tradition.

Knowing fans realize it's not a stand-alone stat. Also, it gains credibility as the volume of decisions rises, even though it's still subject to scrutiny.

[STLCards2]

It makes more sense to simply ask how much control do pitchers have over their win/loss record rather than to analyze any correlation(s). Since they don't have any control over how many runs the team scores and since they don't make out the lineup card or play general manager, all they can control is how much they give up. To say pitchers have control over their W-L record is like saying the Yankees have control over how well the Padres do.

Yes - looking at runs allowed is much better than W-L. Never said otherwise. However, how much they give up on their own merits does have some impact on W-L record - which is why Zach Grienk has a much better W-L record on the Royals than does their #5 starter. To claim otherwise would suggest that pitchers on the same teams would have random W-L records, assuming they recieve similar run support. Not a human alive believs that. Pitchers do have some control over W-L record, just less than they do over a whole lot of other stuff - like all that you mentioned. You are getting "outdatted, not very useful, overrated, not the best, very contingent on other factors, etc." which is true, confused with "absolutely, completely meaningless, useless, and 100% impacted by factors other than pitcher quality", which is not true.

Asd far as Win Shares - I would trust WS over W-L record. I take that back. PRAA, WAR, etc. are surely better than W-L. Never desputed that.

[BiZmaRK]

Nor is it subjective when he loses 1-0 in 15 innings. Two pitching workhorses labored against eachother on a good day. One simply had to lose.
Pitcher + Team].
But it wasn't the pitcher who lost. It was the team which lost. Why should the loss go to the pitcher when it was the team that lost?

[BiZmaRK]

Yes - looking at runs allowed is much better than W-L. Never said otherwise. However, how much they give up on their own merits does have some impact on W-L record - which is why Zach Grienk has a much better W-L record on the Royals than does their #5 starter. To claim otherwise would suggest that pitchers on the same teams would have random W-L records, assuming they recieve similar run support.
If you're going to say the stat is legitimate because the run support is similar, then why not use a league-wide standard for run support over the number of innings he pitches and compare that with the number of runs he allows.

How are win shares calculated?

[STLCards2]

If you're going to say the stat is legitimate because the run support is similar, then why not use a league-wide standard for run support over the number of innings he pitches and compare that with the number of runs he allows.

How are win shares calculated?

Where did I say it was "legit?" I just said it wasn't completely useless.

All one would need to do to see which impacts W-L record the most between run support or ERA, would be to look at the top 50 W-L % leaders and see where they rank in terms of ERA or in terms of run support. If run support is all that matters (as some seem to be suggesting), you should see all of the top W-L guys having the best run support and all of the worst W-L guys getting the worst run support. Likewise for ERA and W-L.

What you will find is some correlation between W-L and quality of pitching (ERA) and some between W-L and amount of run support, but there will be enough guys on poor offensive teams with good W-L records to prove that run support is not the only factor. And you will see enough guys with mediocre ERA's and good records to prove that run support is a major factor.

And there are adjustment/neutralizing tools out there that do adjust run support to be even for everybody. And unlike the "all W-L record is determined by run support" theory would suggest, every pitcher does not end up with the same W-L record.

The issue is - things like WAR, RSAA, WSAB, FIP, and even ERA+ are better than W-L, so why use it? You, me, and curveball are all in agreement here. So you think W-L record in solely dependent on teammates (which would imply that pitcher skill and W-L records are completely non-related. I think they are somewhat related, but not enough-so to use W-L record confidently as an evaluation tool. We are not far apart at all, so this semantics debate is over on my end.

As far as Win Shares - the basics are that each team win is divided evenly to players. Many, many issues with the system (primarily the arbitrary pitcher/defense split in assigning responsibility for run prevention, and a lack of "loss shares"). The ins and outs are complicated and available on Bill Jame's websites and books.

[curveball]

Where did I say it was "legit?" I just said it wasn't completely useless.

All one would need to do to see which impacts W-L record the most between run support or ERA, would be to look at the top 50 W-L % leaders and see where they rank in terms of ERA or in terms of run support. If run support is all that matters (as some seem to be suggesting), you should see all of the top W-L guys having the best run support and all of the worst W-L guys getting the worst run support. Likewise for ERA and W-L.

What you will find is some correlation between W-L and quality of pitching (ERA) and some between W-L and amount of run support, but there will be enough guys on poor offensive teams with good W-L records to prove that run support is not the only factor. And you will see enough guys with mediocre ERA's and good records to prove that run support is a major factor.

And there are adjustment/neutralizing tools out there that do adjust run support to be even for everybody. And unlike the "all W-L record is determined by run support" theory would suggest, every pitcher does not end up with the same W-L record.

The issue is - things like WAR, RSAA, WSAB, FIP, and even ERA+ are better than W-L, so why use it? You, me, and curveball are all in agreement here. So you think W-L record in solely dependent on teammates (which would imply that pitcher skill and W-L records are completely non-related. I think they are somewhat related, but not enough-so to use W-L record confidently as an evaluation tool. We are not far apart at all, so this semantics debate is over on my end.

As far as Win Shares - the basics are that each team win is divided evenly to players. Many, many issues with the system (primarily the arbitrary pitcher/defense split in assigning responsibility for run prevention, and a lack of "loss shares"). The ins and outs are complicated and available on Bill Jame's websites and books.


I know we are all on the same page here, but is it not a bit unfair to assume that a pitcher with an era of 3.00 with run support of 4 runs per game is equal to a pitcher with an era of 5.00 with run support of 4.5 runs per game, and expect the pitcher with the higher era to have a better w/l record?

Wouldn't you have to find a point where two full runs lower of era has to equal x amount of run support extra per start before you can really determine whether era or run support has a greater influence on w/l record?

All I know is that there have been many instances of pitchers with very good eras, yet they have very mediocre w/l records, and the answer is usually poor run support. Nolan Ryan led the league in era+ in 1987, yet had a w/l record of 8-16. That in itself points to how useless w/l records can be. I honestly know of no other stat like w/l record that holds value in some peoples' eyes, yet can be so utterly misleading.

On the other hand, has there ever been a pitcher in MLB history that has ever had a losing record when his team has averaged at least 5 runs for him? Has there ever been a pitcher who finished 8-16 despite having the best run support per game average?

It is from this perspective that I believe, rightly or wrongly, that run support has a greater influence on w/l record than era. In the end, I guess it really doesn't matter because the three of us are close enough in agreeance as to how little value w/l records have.

[STLCards2]

Has there ever been a pitcher who finished 8-16 despite having the best run support per game average?

.

Well, pitchers who are so bad that they would go 8-16 with good run support don't make at least 24 starts per season. Horrible starters (replacement level) would produce pretty poor records on good offensive teams if they were given enough GS's, yes. Of course their records would be even worse on mediocre teams, etc. That is why W-L is poor to look at in isolation. Not because it has 0 value.

The better question to as is: which are there more of: Nolan Ryan seasons (bad W-L record with great ERA on bad offensive team) or more Zach Grienke seasons (good W-L record with great ERA on bad offensive team).

[leewileyfan]

Take Cleveland in 2005. Cliff Lee was 18-5, and Kevin Millwood was 9-11. Is this really a no-brainer because W/L record is so meaningful? I would guess that Millwood would beat Lee in win shares, warp, praa, prar, rsaa, and any other metric that tries to put a value on a pitcher's season. If this is indeed the case, then doesn't that show exactly how useless and misleading W/L record can be, and how those other metrics are still way better than W/L record?

Here you have raised a valid question, which deserves further study. It is not so much a matter of Milwood being > Lee. It was all about batting support behind him.

A search of B-R, advanced pitching statistics, reveals that when Lee started games, CLE average scoring 6.5 runs behind him. In games started by Milwood, CLE scored 3.6 runs.

Lee and Milwood were virtually equal in all key departments. The statistics you cite as proving Milwood > Lee would do no such thing. They would indicate that two CLE aces got very opposite offensive support in games they started. They would show that Milwood was far better than his losing record would indicate, not that Milwood > Lee.

A fair evaluator would likewise add that Lee's 18-5 record that season was proof of any kind that Lee > Milwood, either.

CLE effectively blew the pennant with its batting when Milwood was the starter.

[BiZmaRK]

Where did I say it was "legit?" I just said it wasn't completely useless.

All one would need to do to see which impacts W-L record the most between run support or ERA, would be to look at the top 50 W-L % leaders and see where they rank in terms of ERA or in terms of run support. If run support is all that matters (as some seem to be suggesting), you should see all of the top W-L guys having the best run support and all of the worst W-L guys getting the worst run support. Likewise for ERA and W-L.

What you will find is some correlation between W-L and quality of pitching (ERA) and some between W-L and amount of run support, but there will be enough guys on poor offensive teams with good W-L records to prove that run support is not the only factor. And you will see enough guys with mediocre ERA's and good records to prove that run support is a major factor.

And there are adjustment/neutralizing tools out there that do adjust run support to be even for everybody. And unlike the "all W-L record is determined by run support" theory would suggest, every pitcher does not end up with the same W-L record.

The issue is - things like WAR, RSAA, WSAB, FIP, and even ERA+ are better than W-L, so why use it? You, me, and curveball are all in agreement here. So you think W-L record in solely dependent on teammates (which would imply that pitcher skill and W-L records are completely non-related. I think they are somewhat related, but not enough-so to use W-L record confidently as an evaluation tool. We are not far apart at all, so this semantics debate is over on my end.

As far as Win Shares - the basics are that each team win is divided evenly to players. Many, many issues with the system (primarily the arbitrary pitcher/defense split in assigning responsibility for run prevention, and a lack of "loss shares"). The ins and outs are complicated and available on Bill Jame's websites and books.
If we're in agreement here, then let's move on to the next area of concern: Why so many people in the baseball world believe that pitchers W-L has enough merit for it to be listed among the top pitching statistics.

I can only guess that it is so embedded into the culture and tradition of baseball that it would seem foreign to not pay attention to it.

[AstrosFan]

It makes more sense to simply ask how much control do pitchers have over their win/loss record rather than to analyze any correlation(s). Since they don't have any control over how many runs the team scores and since they don't make out the lineup card or play general manager, all they can control is how much they give up. To say pitchers have control over their W-L record is like saying the Yankees have control over how well the Padres do.


Yes - looking at runs allowed is much better than W-L. Never said otherwise. However, how much they give up on their own merits does have some impact on W-L record - which is why Zach Grienk has a much better W-L record on the Royals than does their #5 starter. To claim otherwise would suggest that pitchers on the same teams would have random W-L records, assuming they recieve similar run support. Not a human alive believs that. Pitchers do have some control over W-L record, just less than they do over a whole lot of other stuff - like all that you mentioned.

This seems quite a distortion, because the argument that the runs a pitcher allows is all he controls inevitably leads to the conclusion that that is the aspect of pitching he controls which can affect his W-L record, and all other influences are beyond his control. It's not an argument that the pitcher has zero control.


Well, pitchers who are so bad that they would go 8-16 with good run support don't make at least 24 starts per season. Horrible starters (replacement level) would produce pretty poor records on good offensive teams if they were given enough GS's, yes. Of course their records would be even worse on mediocre teams, etc. That is why W-L is poor to look at in isolation. Not because it has 0 value.

The better question to as is: which are there more of: Nolan Ryan seasons (bad W-L record with great ERA on bad offensive team) or more Zach Grienke seasons (good W-L record with great ERA on bad offensive team).

Ryan's ERA+ was 142. Greinke's was 205. I think that probably had an effect.

[curveball]

Here you have raised a valid question, which deserves further study. It is not so much a matter of Milwood being > Lee. It was all about batting support behind him.

A search of B-R, advanced pitching statistics, reveals that when Lee started games, CLE average scoring 6.5 runs behind him. In games started by Milwood, CLE scored 3.6 runs.

Lee and Milwood were virtually equal in all key departments. The statistics you cite as proving Milwood > Lee would do no such thing. They would indicate that two CLE aces got very opposite offensive support in games they started. They would show that Milwood was far better than his losing record would indicate, not that Milwood > Lee.

A fair evaluator would likewise add that Lee's 18-5 record that season was proof of any kind that Lee > Milwood, either.

CLE effectively blew the pennant with its batting when Milwood was the starter.


Millwood was better than Lee.

Hitters hit .251/.295/.403/.698 against Lee
Hitters hit .248/.300/.387/.688 against Millwood

Lee had an era of 3.79 and an era+ of 111
Millwood had an era of 2.86 and an era+ of 146

I am pretty sure that Millwood beats Lee in win shares, warp, war, rsaa, praa, praa, and any other metric out there. This makes perfect sense. It follows logic that Millwood should beat Lee because he pitched better that year.

The only thing that doesn't make sense is that Lee finished 18-5, and Millwood finished 9-11, which is why I, and many others put little to no stock in w/l records.

[AstrosFan]

What causes Millwood's ERA+ to be so much better than Lee's, I wonder?

M vs. L

WHIP: 1.219 to 1.218.
H/9: 8.5 to 8.6.
BB/9: 2.4 to 2.3
K/9: 6.8 to 6.4
HR/9: .9 to 1.0
K/BB: 2.81 to 2.75
BABIP: .284 to .279

Really not the stark differences one would expect in a 35 point difference in ERA+.

Since OBP is more important than SLG, the edge Millwood has is smaller than the .688 to .698 difference suggests.

Millwood allowed 33 SB, 6 CS. Lee was at 7 and 4.

Millwood induced 19 GDP, Lee 10.

There are some factors I'm leaving out, but I'm struggling to see how the ERA+ difference was so large.

[STLCards2]

There are some factors I'm leaving out, but I'm struggling to see how the ERA+ difference was so large.

LOB% maybe?

Also: "Ryan's ERA+ was 142. Greinke's was 205. I think that probably had an effect".

I selected these two only because both have been mentioned in this thread so far. I guess I could have provided parameters for pitchers with "bad W-L record with great ERA on bad offensive team" or "good W-L record with great ERA on bad offensive team" by picking specific levels of run support/ERA, etc, but I think everybody got the main point.

[STLCards2]

The only thing that doesn't make sense is that Lee finished 18-5, and Millwood finished 9-11, which is why I, and many others put little to no stock in w/l records.

Run support is not evenly distributed for each pitcher. Take Carlton in 1972, for example. His team's overall offense was terrible, but if I am not mistaken, he actually got pretty good run support personally. Now we can have a debate all day about if good pitchers somehow "inspire" better offense, but that hasn't been proven, if it does happen.

Also remember that runs allowed aren't distributed evenly either. This is another reason outside of run support why a guy's W-L record may not be perfectly representative of how he pitched. Not to mentioned defensive support, which has seldom been mentioned yet.

So for Milwood vs. Lee, it would be helpful to see how their RAs were distributed and how much run support each received individually - not just how good or bad their team's offense is.

[AstrosFan]

LOB% maybe?

79.1 to 72.0 edge for Millwood. Really don't know how much an effect it would have.

[STLCards2]

79.1 to 72.0 edge for Millwood. Really don't know how much an effect it would have.

Millwood was 2nd in the AL (typical league leader is near 80%) and Lee was at about league average (around 71%). Seems like this may be a pretty big (if not the biggest) factor in the ERA+ gap.

[curveball]

Run support is not evenly distributed for each pitcher. Take Carlton in 1972, for example. His team's overall offense was terrible, but if I am not mistaken, he actually got pretty good run support personally. Now we can have a debate all day about if good pitchers somehow "inspire" better offense, but that hasn't been proven, if it does happen.

Also remember that runs allowed aren't distributed evenly either. This is another reason outside of run support why a guy's W-L record may not be perfectly representative of how he pitched. Not to mentioned defensive support, which has seldom been mentioned yet.

So for Milwood vs. Lee, it would be helpful to see how their RAs were distributed and how much run support each received individually - not just how good or bad their team's offense is.


Isn't that an argument as to why w/l records can be grossly misleading, and therefore pretty much useless?

Era+, win shares, war, warp, praa, prar, etc...., will all have Millwood beating Lee. The only measure that has Lee beating, actually crushing Millwood is w/l record. Therefore, one can come to their own conclusions as to the merits of w/l record.

It seems like w/l record comes with a lot of fine print.

WARNING

** W/L records are only somewhat useful when pitchers have similar eras and run support. In cases where either era, run support, or both, are not similar, then w/l records are not really useful, and we assume no liability **

[STLCards2]

Isn't that an argument as to why w/l records can be grossly misleading, and therefore pretty much useless?



Not only does it support the shakiness of using W/L records, but I even flat-out stated such in the very post you quoted...

"This is another reason outside of run support why a guy's W-L record may not be perfectly representative of how he pitched."

[curveball]

I would like to hear from anyone who really believes that w/l records are useful.


CLEVELAND 2005

Cliff Lee*****Kevin Millwood

ERA 3.79*****2.86
ERA+ 111*****146

RAA 10*****21
PRAR 29*****39
PRAA 7*****19
DERA 4.17*****3.53
WARP3 3.6*****5.2
PRAR1 34*****43

W/L 18-5 .783*****9-11 .450


The only stat that Lee beats (annihilates) Millwood is W/L %. Could it be that all the other stats are wrong or misleading? Or isn't it most likely that there is an enormous problem with W/L records?

I know of no other stat that is so flawed as w/l records, yet so many people use it as a main point in their arguments. Also, if there is an example of any other stat used to evaluate pitchers that can be so grossly misleading or useless like the w/l example above, I would like to know which one it is.

From the example above, can you really blame anybody for believing that w/l records can be so unbelievably bad in evaluating pitchers that they just pretty much ignore them completely?

Hell, I'd be pretty embarrassed if I invented w/l records and tried to pass it off as some accurate representation of how well a pitcher pitched. We actually have people who champion w/l records, and shun better evaluation tools like the ones presented above. :banghead:

As flawed as win shares, or warp3, etc... may be, if there is an example where a pitcher A had 40 win shares, and pitcher B had 15 win shares, or pitcher A had a warp3 of 8.1 and pitcher B had a warp3 of 3.5, and it can be shown that pitcher B was actually better than pitcher A, then I'll lose faith in those two measures. No other stats used to evaluate pitchers has ever been shown to be so far off base than w/l records.

[brett]

Maybe a good question is:
"What factors WITHIN A PITCHER'S CONTROL can make his W-L record better or worse than ERA would indicate?"

versus factors probably out of his control like most likely run support.

Being able to go more innings could make a difference.
Run distribution.
Pitching differently in closer games.
Having more earned runs-say blowing it after the 3rd out should have been made?

I mean, if Grienke can really turn it on with runners on base, which he definitely appeared to, he should also be able to do so when the score is close.

[leewileyfan]

I took about 2 hours to check the 2005 game logs for both Lee and Milwood [CLE, 2005] and found some revealing stuff:

1. Both Millwood and Lee were subject to the decisions of a manager tending to use a quick hook, tending to go to the pen regardless of how well his starters were doing.

2. Millwood was yanked after six innings on April 2, with the following line;
[all subsequent lines will be presented in the same format]:

IP............H............R..........ER.......... .Millwood.............CLE

6.0.........4..............0..........0........... ....ND.............Lost, 4-3

3. The pendulum slid slightly the other way on May 4 and again on May 14, when he was yanked with these lines:

5.0.......11..............4..........3............ ..W..............Won, 5-4

3.0........5...............2..........2........... ....ND............Won, 3-2

4. After being out for 21 days, Millwood had four consecutive starts in which he seems to have been under close watch, with these three lines for June 16, June 21 and June 27 respectively:

5.0........4..............3...........1........... ....ND............Won, 5-4

6.0........7..............6...........5........... ....L...............Lost, 9-2

6.0........3..............0...........0........... ....W..............Won, 7-0

July 14 was a heartbreaker:

8.0........6..............1...........1........... ....L...............Lost, 1-0

6. On August 4, August 17, August 27, September 3, September 8, September 13, September 19, September 24 and September 30, Millwood had 10 solid starts, with the following cumulative numbers:

70.0......69............21.........16............. 4-2 + 4 ND...CLE: 5-5

7. Luck broke Millwood's way on August 22:

6.0.........5..............4..........4........... ...W...............Won, 11-4

8. Lee had a rocky April 7 opener, his bacon saved by later-in-game team offense:

3.3..........8.............5...........5.......... ...ND..............Won, 11-5

His next start on April 13 was better and resulted in a "fair enough" ND:

6.7..........6............4............4.......... ...ND..............Lost, 5-4

May 6 Lee had his bacon saved again:

5.3..........8............3............3.......... ...W...............Won, 8-6

9. However, Lee, too was visited by the fickle finger of fate, on June 7:

7.0..........5............0.............0......... ....ND.............Won 5-3

..... and again, on June 22:

7.0..........6............2.............2......... ....ND............Lost, 5-4

..... and again August 14:

8.0..........6............0.............0......... ....ND............Lost, 1-0

..... August 31:

6.0..........5............2.............2......... ....ND............Lost, 4-3

...... September 29:

8.0..........6...........1..............1......... ....L..............Lost, 1-0

Wrapping up with Lee, he has bacon-saving team offensive bailouts that resulted in wins for shaky starts, by date:

June 17 [13-6]; August 25; [12-4] and September 22 [11-6].

Anyone taking time to read this and pursue it further [Ref: Baseball-Reference; Pitching; Advanced, Game Logs may find agreement with the following observations:

-tendency to yank starters early;
-erratic bullpen results at least for these two starters on the CLE staff;

-a portrait of two pitchers, very nearly identical on all key internal pitching elements, experiencing diametrically opposed W-L outcomes.

I do no cite LOB% for these pitchers, because in the context of this discussion, I believe I would be mis-applying the stat.

However, if someone can find CLE 2005 bullpen pitching splits, I'd be willing to bet that they allowed a high rate of inherited runners to score, especially when relieving either Millwood or Lee ..... probably worse results with Lee's inherited runners [Thereby closing the ERA and ERA+ gap].

[digglahhh]

What do you have against semantics, which you seem to consider an epithet?

Nothing, Korzybski is my homeboy. In fact, I'd consider metalinguistics among my favorite subjects. My point in making a plea to remove ourselves from the semantic weeds is that the spirit of this debate was a bit undermined when the discussion got bogged down in differentiating between something being worthless as opposed to simply being of comparatively low value, or worth.

In the spirit of this discussion, "worthless" and "often, but not always, indicative--and only as a thumbnail, at that" is pretty much a false distinction. Nobody argues that the stat is valueless, just of relatively weak value as compared to a wealth of other data that is just as easily and widely available.

So, for anybody who is arguing simply that W-L is not entirely worthless, they have triumped gloriously in an exclusively academic debate against adversaries who either don't exist, or only appeared so because they used specific words too liberally, taking for granted the context of a thread located within the Statisticss section of a baseball message board.

Would anybody now like to make the claim that W-L is a particularly valuable stat in comparison to and independent of readily available alternatives?


Yep. Let me start with four, just for openers:

1. Marcels
2. Win Shares
3. FB/GB/LD%
4. BABIP

Would you care to share your reasoning as to why W-L is a better (interpret that as you wish) stat than those you've chosen? I'm not going to flip out hailing win shares at some divine decree or anything, just curious as to why you reached the conclusion you did.

...Frankly, I could agree on whatever case you make to dismiss all of them and it wouldn't impact the veracity my initial point one iota.

[leewileyfan]

Nothing, Korzybski is my homeboy. In fact, I'd consider metalinguistics among my favorite subjects. My point in making a plea to remove ourselves from the semantic weeds is that the spirit of this debate was a bit undermined when the discussion got bogged down in differentiating between something being worthless as opposed to simply being of comparatively low value, or worth.

I am not about to get bogged down in a debate on semantics. Your homeboy may be Korzybski whose view of semantics was a specialized measurement of the scientific, natural limitations imposed on the human organism .... that physical stuff of which our senses are composed. I'm more in the Russell, Saussiere, Eco and Chomsky camp that focuses a bit more of human interpretation, application and motivation via the perceptions we are capable of making.



Would you care to share your reasoning as to why W-L is a better (interpret that as you wish) stat than those you've chosen.

I'm guessing you mean the ones I've chosen to eliminate. I don't have to share any reasoning for why I believe W-L is better than they, simply because I did not make reference to a pecking order of significance for either the ones I retaines, W-L, or the ones I dumped.


...Frankly, I could agree on whatever case you make to dismiss all of them and it wouldn't impact the veracity my initial point one iota.

You seem to believe I'm championing something here. I am not.

Part of my acceptance of W-L as having some value may well be a matter of imprinting: It was a big deal when I was a kid in the formative years of my fanhood. It certainly was an eye-catcher in the morning papers, with morning lines like Feller [12-7] at Gomez [11-4].

However, it didn't take too long to realize that Leonard [7-9] at Chandler [10-4] would make for an equally exciting match-up. Fans interested enough to dig deeper want to see raw numbers in terms of larger context, the dynamics that can drastically modify first impressions.

Markhov provides a mathematical construct for crystal-gazing into a projected career path. BABIP evaluates safe hits as a percentage of batted balls, strikeouts excluded. Too much random chance there. for me. I'd rather dive deeper into constructing a defense metric that goes beyond BABIP.

FB, GB, LD%%% are useless to me, especially since, as recently as two years ago, spotters feeding data to sabermetric model makers were quite at odds as to what they individually identified as line-drives.

ERA+ simply tells me what the quotient is when I divide LG ERA by an individual pitcher's ERA. I want to know how relievers performed with inherited men on base from a starter.

Was starter A charged with 17 ER scored of 34 men left on base, that relievers allowed to score? Was starter B, on the same team, more fortunate, in that, of 33 runners he left on, the 'pen allowed only 9 to score?

The longer the career the more support one can credit to W-L, due to volume of decisions. On a season-to-season basis, it's a neon light to draw attention to pitchers to explore the relationship between Wins and effectiveness.

We have largely been talking about starters. How about HoF closers with losing records? There's a whole new thread buried in there.

Some will argue that closers should never be judged in any way by W-L records, since they arrive in high leverage situations with the cards stacked against them in every way.

However, a closer who comes in in a total jam situation, is off the hook for 3 runs when he inherits a bases-loaded situation.

Just using that as an example, say Hillenbrand von Schtupt is the reliever, bases full, team ahead 10-7 in the eighth. He allows all three runs to score before fanning the side, stranding a runner on 2B. He has this line:

IP................R................ER............H ..............BB............K

1.0..............0.................0.............1 ...............0............3

ERA = hero. ERA+ = hero. Inherited runners score % = not so much. How does von Schtupt stand to lose? He yields a run in the ninth, losing 11-10. His team ties in the 9th, 11-11 and he loses in the 10th.

It's not W-L that is weak [as constituted in current rulemaking]. It's the mis-application of the record without digging deeper. Even then, some of the finer individual stats provide only part of the context.

Hmmmmmm, context! Now we're back to semantics.

[STLCards2]

So, for anybody who is arguing simply that W-L is not entirely worthless, they have triumped gloriously in an exclusively academic debate against adversaries who either don't exist, or only appeared so because they used specific words too liberally, taking for granted the context of a thread located within the Statisticss section of a baseball message board.

.

Hey, at least I trumphed gloriously at something in life! Maybe next time they should not use specific words so liberally.:happy:

[BiZmaRK]

What causes Millwood's ERA+ to be so much better than Lee's, I wonder?

M vs. L

WHIP: 1.219 to 1.218.
H/9: 8.5 to 8.6.
BB/9: 2.4 to 2.3
K/9: 6.8 to 6.4
HR/9: .9 to 1.0
K/BB: 2.81 to 2.75
BABIP: .284 to .279

Really not the stark differences one would expect in a 35 point difference in ERA+.

Since OBP is more important than SLG, the edge Millwood has is smaller than the .688 to .698 difference suggests.

Millwood allowed 33 SB, 6 CS. Lee was at 7 and 4.

Millwood induced 19 GDP, Lee 10.

There are some factors I'm leaving out, but I'm struggling to see how the ERA+ difference was so large.
Perhaps one pitcher had a tendency to space out the runners he allowed on base while the other had a tendency to allow base runners in bunches.

A good stat to compare is what is the oppositions OBP when there are runners on base.

[leewileyfan]

OK, here's one clue. A review of relief pitcher game logs for the CLE 2005 bullpen reveals the following numbers:

Games Started by..............Inherited Runners.........Inherited Runners Score

Lee, April.....................................6....... ...........................2
Millwood, April...............................4............. .....................1

Lee, May......................................8........ ..........................2
Millwood, May................................3.............. ....................0

Lee, June.....................................6........ ...........................5
Millwood, June...............................1.............. .....................0

Lee, July.......................................5...... ............................2
Millwood, July................................7............. .....................3

Lee, August..................................6......... ..........................1
Millwood, August............................0............... ....................0

Lee, September.............................0........... ........................0
Millwood, September.......................9................. ..................2

October = 0 for both pitchers.

Totals [2005] Lee ......................31.......................... ..12 [38.7%]

.....................Millwood.................24.. ............................6 [25.0%]

That's one modifier.

[BiZmaRK]

OK, here's one clue. A review of relief pitcher game logs for the CLE 2005 bullpen reveals the following numbers:

Games Started by..............Inherited Runners.........Inherited Runners Score

Lee, April.....................................6....... ...........................2
Millwood, April...............................4............. .....................1

Lee, May......................................8........ ..........................2
Millwood, May................................3.............. ....................0

Lee, June.....................................6........ ...........................5
Millwood, June...............................1.............. .....................0

Lee, July.......................................5...... ............................2
Millwood, July................................7............. .....................3

Lee, August..................................6......... ..........................1
Millwood, August............................0............... ....................0

Lee, September.............................0........... ........................0
Millwood, September.......................9................. ..................2

October = 0 for both pitchers.

Totals [2005] Lee ......................31.......................... ..12 [38.7%]

.....................Millwood.................24.. ............................6 [25.0%]

That's one modifier.
I strongly recommend you read Baseball Between The Numbers. When you do, you'll see that all these "inherited runners scoring" calculations you've made takes you down an unnecessary path. BBTN suggests you not go by the actual number of runners scored and go by the league average for what scores based on # of runners on base and # of outs.

For example, let's say the league average runs scored when there is 1 out and a runner on third is .90. A pitcher comes out of the game with 1 out and a runner on third. He is charged .90 runs - whether or not that run scores. Same principal for relief pitchers entering the game. Enter the game in that same situation and you're charged .90 runs. If none score, then you get an ERA of -.90. If one run scores, you get .10 runs charged to you. And as you can see, the .10 charged to the reliever and the .90 charged to the starter add up to 1.0 - precisely the number of runs actually allowed.

[leewileyfan]

I strongly recommend you read Baseball Between The Numbers. When you do, you'll see that all these "inherited runners scoring" calculations you've made takes you down an unnecessary path. BBTN suggests you not go by the actual number of runners scored and go by the league average for what scores based on # of runners on base and # of outs.

I get a big kick when such strongly recommed-ed proposals are made, presuming the recipient is unfamiliar with the work being offered for his enlightenment.


For example, let's say the league average runs scored when there is 1 out and a runner on third is .90. A pitcher comes out of the game with 1 out and a runner on third. He is charged .90 runs - whether or not that run scores. Same principal for relief pitchers entering the game. Enter the game in that same situation and you're charged .90 runs. If none score, then you get an ERA of -.90. If one run scores, you get .10 runs charged to you. And as you can see, the .10 charged to the reliever and the .90 charged to the starter add up to 1.0 - precisely the number of runs actually allowed.

Sure, go for the mathematical model when real numbers are available to be ignored. Isn't that something similar to brilliant trading derivatives that brought down a chunk of Wall Street?

IMHO, the very best stat to trace pitchers in the accumulation of 24 base-out situation befores & afters for each pitcher's entry and leaving of games [unembellished with added mathematical augmentation].

[leewileyfan]

Having read and re-read all posts here, scouting various sites on the 'net, and jotting notes of my own, I am satisfied [for me, not trying to recruit converts, that I have found the ultimate stat for the following much-debated player performance issues:

-merit-based pitcher deserved wins that takes the actual W-L record and holds it up for comparison in in game situational effectiveness over the season;

-relief pitcher effectiveness in real-game situational performance, similarly held up to ERA, ERA+, W-L, saves, and holds recorded;

-the ultimate grand-scale clutch performance of batters over an entire season, based solely on actual situations not hand [cherry] picked leverage situations;

Up front, I anticipate some sabermetric confrontation on this, the essential counter-arguments based on a " ..... but that's not what thise stat was devised for."

My answer can only be the analogy with familiar scientific discovery stories or aha moments that arose historically when a lab technician left some materials out over night, when protocol called for special storage ..... serendipity.

The stats I'm idendifying are:

WPA: Win Probability Added, inspired by the 24 base-out situations graphic that puts a run probability alue to each of the 24 identified situations.
It's a before-and-after situation:

Example: Pitcher A, the starter, went five scoreless innings but ran into trouble in the sixth, giving up 5 runs, the last on a triple, after one opponent had been retired.

Right now, the 5 runs allowed = 5, against pitcher A. However, WPA pencils in and additional .897, because, after millions of simulations, that is the run expectancy for the base-out situation identified as: runner on 3B; one out.

Pitcher B, A's reliever, enters with a runner on 3B, not his personal liability. However, the .897, presently penciled in as a possible added liability for A, hangs overhead because the game situation is now in a continuum.

Pitcher B fans the first batter he faces. The situation is now runner on 3B, with 2 out = .382 on the run probability graphic. Relief pitcher B is now [temporarily] in the black, by .897 - .382 = .515, because his out lessened the threat to his team by that amount. NOTE: Pitcher A is still on the hook for .897, because that's the situation he left behind.

If Pitcher B retires the next batter and the side is out without that runner scoring from 3B, both pitchers benefit:

Pitcher A is logged in with 5 on his ledger, his .897 is expunged.

Pitcher B is logged in with -0.897, since he reduced a liability inherited to 0.

However, if pitcher B, after getting the second out, yields a single and the runner scores, the lines look like this:

Pitcher A = 5 + .897 = 5.897
Pitcher B = .103, since .897 + .103 = the shared responsibility for that 6th run scoring.

Pitcher B would also now be in his own situation @ +.209, his runner at 1B, with 2 outs. His .209 + .103 = .312.*

*The Hidden Game of Baseball; p. 153; Thorn, John & Palmer, Pete

WPA & RE24 & REW are related stats that take these situations, compile them for games and then for full seasons, culminating in RE24 being converted to "Wins" [REW].

The same can be done for hitting. All situations enter the mix; and those disposed to identifying and narrowing clutch situations may do so; but WPA and RE 24 do the +/- for all PA, which may be simpler and more revealing of offensive value.

REF: These statistics are available in detail @ FanGraphs. Click batting or pitching; then click on Win Probability options.

Relative to Lee and Millwood and CLE 2005, WPA, RE24 and REW compare them as follows:

Pitcher...............WPA..............RE24....... ...REW...........[E] Wins

Lee....................11.09............1.72...... .....1.22............12.22

Millwood..............22.37............1.82....... ....2.25.............12.25

As I mentioned earlier, the two pitchers are virtually identical in pitcher-centric efficiencies; but Millwood's superior WPA is wasted due to poor batting support.

When Millwood started, CL averaged @ 3.6 runs scored behind him. With Lee, CLE's offense averaged 6.5 runs scored.

[digglahhh]

LWF,

I do not feel you are championing anything radical in regards to the "value" of the win as a stat. And, I agree that W-L, like many other stats are vastly more useful in context.

Two quick points though. One, many other stats, even something as simple as ERA+ subsumes far more context (by virtue of its inputs) than W-L does. So, even by the measuring stick of "context" even simple stats such as ERA+ outperform W-L, to say nothing of far more advanced stats like FIP. As for stats like trajectory percentages, these are empirical and therefore valuable in the sense that they can be components of, and oversight for other stats. For example, if you know BABIP and trajectory %, you may be able to hint at the reliability of the BABIP number. W-L is more of a value judgment stat, that can subjectively assign different interpretations to qualitatively equal performances. And, FTR, I share your skepticism about the accuracy of the trajectory recordings. I have intimate knowledge of many of these errors; I think some of the vets of this section will attest that I've complained about the accuracy of this type of data and warned against taking it as gospel many times in the past.

Two, more context is always better. Better to scour game-by-game pitching lines, even better to watch all of the games to identify individual contexts and dynamics - dubious umpiring calls which affect stats, etc. But, at the same time, the purpose of stats are to provide a thumbnail. A good stat will be able to pack a lot of context into that thumbnail, but nothing is perfect. I acknowledge that any knowledge/context represented by a stat is partial. But when trying to get a beat on broad bands of players across history, the stat, though reductionist by nature, is a necessary evil. It is simply impossible to investigate the most nuanced context behind stats for tons and tons of players. So, at some point, as a matter of practicality, we all settle on a stat, or a set thereof, that we use as a rough de facto indicator of a pitcher's season (or career). When choosing that barometer, wins is a poor choice in my opinion. That's all my argument really is.

And, just to make one other point, if part of the value of wins (to you) is the sentimentality of them, imprinting, and the role the W-L played in fostering your love of the game, then that is a real personal value that stat has for you. So, I won't deny that. It is arbitrary and second-hand, however. Had the powers that be chose a different form by which to define the match-ups and rivalries among pitchers, you'd have developed some relationship with that other stat just as easily.

[leewileyfan]

LWF,And, just to make one other point, if part of the value of wins (to you) is the sentimentality of them, imprinting, and the role the W-L played in fostering your love of the game, then that is a real personal value that stat has for you. So, I won't deny that. It is arbitrary and second-hand, however. Had the powers that be chose a different form by which to define the match-ups and rivalries among pitchers, you'd have developed some relationship with that other stat just as easily.

Correct. Which is why I so readily used the word imprinting on my own generation.

We basically agree on salient points. In fact, all statistics must be presented in context to earn credibility.

If I say Sherry Robertson [just to pull a real player name out of thin air] was the greatest hitter of all time because he hit .600, the men with the nets would be justified in seeking me out for questioning. If Sherry went 3 for 5 in one season and I presented that as the whole of his worth, I'd be daft.

W-L gains credibility with volume of numbers, which is further driven by career longevity and rate of work. I fully realize its origins as a drawing card to the ballpark, a marquee item. However, I am seduced by the allure of a 20 - game winner, a fading class; and I can respect the accomplishment even if closer inspection proves that some 12-14 guy was actually a better pitcher based on the nuts & bolts of the bigger picture, just a guy with worse support and poorer luck.

In the CLE 2005 debate, I can fully empathize with Millwood fans who decry his 9-11 record. However, I am not inclined to diminish Lee's capabilities in compiling an 18-5 record for the same club.

As long as any fan is willing to dig behind the glaring numbers, there's no need to demolish the W-L promotional headline grabber. If nothing else, it suggests guys who stayed longer; kept closer; OR held on until his team could pull rabbits out of hats for him.

[curveball]

Correct. Which is why I so readily used the word imprinting on my own generation.

We basically agree on salient points. In fact, all statistics must be presented in context to earn credibility.

If I say Sherry Robertson [just to pull a real player name out of thin air] was the greatest hitter of all time because he hit .600, the men with the nets would be justified in seeking me out for questioning. If Sherry went 3 for 5 in one season and I presented that as the whole of his worth, I'd be daft.

W-L gains credibility with volume of numbers, which is further driven by career longevity and rate of work. I fully realize its origins as a drawing card to the ballpark, a marquee item. However, I am seduced by the allure of a 20 - game winner, a fading class; and I can respect the accomplishment even if closer inspection proves that some 12-14 guy was actually a better pitcher based on the nuts & bolts of the bigger picture, just a guy with worse support and poorer luck.

In the CLE 2005 debate, I can fully empathize with Millwood fans who decry his 9-11 record. However, I am not inclined to diminish Lee's capabilities in compiling an 18-5 record for the same club.

As long as any fan is willing to dig behind the glaring numbers, there's no need to demolish the W-L promotional headline grabber. If nothing else, it suggests guys who stayed longer; kept closer; OR held on until his team could pull rabbits out of hats for him.


Does W/L really gain credibility with volume of numbers? W/L can be so blatantly misleading for a single season because it does not level the playing field for run support differentials. For every season, there will always be a fairly substantial margin of error.

It seems that you are arguing the W/L will overestimate wins some years, and underestimate wins some other years, so in the end it should all even out.

That is like saying something isn't very accurate, but let's use it over the long haul anyways, and hopefully it will work better. Shouldn't the question be, why bother using something that isn't accurate in the first place?

You are honest enough to admit that you have an affinity for wins because of imprinting. You are also knowledgeable enough to know about the serious flaws of w/l records. You clearly display a willingness to discover flaws and find ways to correct them. I think it is time that you try to sever your attachment to wins. It seems like you kind of want to. ;)

You say things like "However, I am seduced by the allure of a 20 - game winner, a fading class; and I can respect the accomplishment even if closer inspection proves that some 12-14 guy was actually a better pitcher based on the nuts & bolts of the bigger picture, just a guy with worse support and poorer luck."

That is kind of a contradiction. You should be respecting the guy who finished 12-14 because he deserved better, and he'll be the one nobody thinks twice about because a lot of people will dismiss him because of his w/l record. If there was no such thing as w/l records for starting pitchers, then people may actually see pitchers for what they really are.

You said it yourself, w/l records are a distraction from the nuts and bolts of the bigger picture.

[curveball]

Since we are talking about wins, what is the best stat, tool, or website that tries to factor in run support, defense, etc...., to at least try to make w/l more meaningful?

I know that baseball-reference has neutralized stats, but it seems that their calculations have changed quite considerably ever since they re-designed their website, and I am not sure if it is for better or worse.

LWF mentioned fangraphs, and I saw that FIP can be translated into wins, but I am not sure how to do it, or how accurate the more knowledgeable posters here think it is.

I am not a fan of w/l, but it is so entrenched in baseball history, that you really can't get away from some people using it in their arguments. I just want to see how actual w/l records deviate from revised w/l records that try to level the playing field by at least factoring in run support.

[leewileyfan]

Does W/L really gain credibility with volume of numbers? W/L can be so blatantly misleading for a single season because it does not level the playing field for run support differentials. For every season, there will always be a fairly substantial margin of error.

What I intended to convey was that W-L gain a certain degree of credibility with longer careers as the W-L numbers accumulate. I can champion a pitcher like "Dutch" Leonard who won almost 200 games with cellar-dwellers while managing to maintain a > .500 W-L %. His 191 wins gives him degree of cachet. even among those who don't respect his career as much as I do.


It seems that you are arguing the W/L will overestimate wins some years, and underestimate wins some other years, so in the end it should all even out.

Yep. I have repeatedly agreed that W-L can present a distorted picture of actual effectiveness. I never presume it all evens out. However, for me, a W-L record can draw my attention whether is is outstanding or awful One guy @ 1940 is 10-1; another @ 1953 is 2-17: I want to look into those record more .... curiosity.


That is like saying something isn't very accurate, but let's use it over the long haul anyways, and hopefully it will work better. Shouldn't the question be, why bother using something that isn't accurate in the first place?

Not literally. I guess my admitted imprinting on W-L is related to its function as an eye-catcher, to draw my attention to a pitcher's performance.

When I see a pitcher's record = 155-187, I know he lasted a long time, so I'm drawn to his career. Similarly, a 34 year old who's done @ 65-56 will catch my eye because he was a winner with maybe an abrupt ending. I see that Bob Feller won fewer than 300 games, I know [as one who grew up with him in his prime] that he lost almost 5 years to WW II. A young kid today, as curious as I, might be drawn to his numbers, wanting to learn more.


You are honest enough to admit that you have an affinity for wins because of imprinting. You are also knowledgeable enough to know about the serious flaws of w/l records. You clearly display a willingness to discover flaws and find ways to correct them. I think it is time that you try to sever your attachment to wins. It seems like you kind of want to. ;)

Thanks for the kind words and observations, really. I guess that, at a time when society at large is disinclined to preserve traditions, I'm slow to get on the dumping barge. It's a little like the 70+ year old Brit, who on the one hand sees ROYALTY as a bunch of empty pomp, circumstance, ritual & economic waste .................... but, on the other hand figures, look at politics [and politicians elsewhere] ............. concluding "What's the harm?"


You say things like "However, I am seduced by the allure of a 20 - game winner, a fading class; and I can respect the accomplishment even if closer inspection proves that some 12-14 guy was actually a better pitcher based on the nuts & bolts of the bigger picture, just a guy with worse support and poorer luck."

That is kind of a contradiction. You should be respecting the guy who finished 12-14 because he deserved better, and he'll be the one nobody thinks twice about because a lot of people will dismiss him because of his w/l record. If there was no such thing as w/l records for starting pitchers, then people may actually see pitchers for what they really are.

That's just it/ I DO respect the guy @ 12-14.


You said it yourself, w/l records are a distraction from the nuts and bolts of the bigger picture.

Yep. If you let them be, without pursuing the matter more deeply.

[curveball]

Having read and re-read all posts here, scouting various sites on the 'net, and jotting notes of my own, I am satisfied [for me, not trying to recruit converts, that I have found the ultimate stat for the following much-debated player performance issues:

-merit-based pitcher deserved wins that takes the actual W-L record and holds it up for comparison in in game situational effectiveness over the season;

-relief pitcher effectiveness in real-game situational performance, similarly held up to ERA, ERA+, W-L, saves, and holds recorded;

-the ultimate grand-scale clutch performance of batters over an entire season, based solely on actual situations not hand [cherry] picked leverage situations;

Up front, I anticipate some sabermetric confrontation on this, the essential counter-arguments based on a " ..... but that's not what thise stat was devised for."

My answer can only be the analogy with familiar scientific discovery stories or aha moments that arose historically when a lab technician left some materials out over night, when protocol called for special storage ..... serendipity.

The stats I'm idendifying are:

WPA: Win Probability Added, inspired by the 24 base-out situations graphic that puts a run probability alue to each of the 24 identified situations.
It's a before-and-after situation:

Example: Pitcher A, the starter, went five scoreless innings but ran into trouble in the sixth, giving up 5 runs, the last on a triple, after one opponent had been retired.

Right now, the 5 runs allowed = 5, against pitcher A. However, WPA pencils in and additional .897, because, after millions of simulations, that is the run expectancy for the base-out situation identified as: runner on 3B; one out.

Pitcher B, A's reliever, enters with a runner on 3B, not his personal liability. However, the .897, presently penciled in as a possible added liability for A, hangs overhead because the game situation is now in a continuum.

Pitcher B fans the first batter he faces. The situation is now runner on 3B, with 2 out = .382 on the run probability graphic. Relief pitcher B is now [temporarily] in the black, by .897 - .382 = .515, because his out lessened the threat to his team by that amount. NOTE: Pitcher A is still on the hook for .897, because that's the situation he left behind.

If Pitcher B retires the next batter and the side is out without that runner scoring from 3B, both pitchers benefit:

Pitcher A is logged in with 5 on his ledger, his .897 is expunged.

Pitcher B is logged in with -0.897, since he reduced a liability inherited to 0.

However, if pitcher B, after getting the second out, yields a single and the runner scores, the lines look like this:

Pitcher A = 5 + .897 = 5.897
Pitcher B = .103, since .897 + .103 = the shared responsibility for that 6th run scoring.

Pitcher B would also now be in his own situation @ +.209, his runner at 1B, with 2 outs. His .209 + .103 = .312.*

*The Hidden Game of Baseball; p. 153; Thorn, John & Palmer, Pete

WPA & RE24 & REW are related stats that take these situations, compile them for games and then for full seasons, culminating in RE24 being converted to "Wins" [REW].

The same can be done for hitting. All situations enter the mix; and those disposed to identifying and narrowing clutch situations may do so; but WPA and RE 24 do the +/- for all PA, which may be simpler and more revealing of offensive value.

REF: These statistics are available in detail @ FanGraphs. Click batting or pitching; then click on Win Probability options.

Relative to Lee and Millwood and CLE 2005, WPA, RE24 and REW compare them as follows:

Pitcher...............WPA..............RE24....... ...REW...........[E] Wins

Lee....................11.09............1.72...... .....1.22............12.22

Millwood..............22.37............1.82....... ....2.25.............12.25

As I mentioned earlier, the two pitchers are virtually identical in pitcher-centric efficiencies; but Millwood's superior WPA is wasted due to poor batting support.

When Millwood started, CL averaged @ 3.6 runs scored behind him. With Lee, CLE's offense averaged 6.5 runs scored.



Does that mean that there is an 89.7% chance of a man scoring from third with one out? That just seems kind of high to me.

The other thing is that I would think it would be better to just charge that .897 to the starting pitcher that left if that is indeed the norm. When that can be expunged, then the starter can either benefit or lose out depending on how good the relief pitching was that came in to mop things up for them. If I understand it correctly, that means that if a starter left with that situation 5 times, and all 5 times he was bailed out by his relievers, then he saves 5 x .897 runs. Whereas if relievers fail all 5 times, then the starter gets charged 5 x .897 runs. So aren't you penalizing starters for having lousy relievers, or who were just unlucky that their relievers failed them? Wouldn't it be better to automatically just charge the starters those runs based on the chances that those runners he left were to actually score? This way, it isn't dependent on the quality of the relief performance.

[leewileyfan]

Does that mean that there is an 89.7% chance of a man scoring from third with one out? That just seems kind of high to me.

Yes. The 89.7% is a percentage conversion of the decimal .897 cited for the base out situation, "B; 1 Out" in the Hidden Game of Baseball; p. 153.

I deliberately selected this particular reference because it is readily accessible for reference, although strict sabermetricians modify the grid each season for "current exactitude."


The other thing is that I would think it would be better to just charge that .897 to the starting pitcher that left if that is indeed the norm. When that can be expunged, then the starter can either benefit or lose out depending on how good the relief pitching was that came in to mop things up for them. If I understand it correctly, that means that if a starter left with that situation 5 times, and all 5 times he was bailed out by his relievers, then he saves 5 x .897 runs. Whereas if relievers fail all 5 times, then the starter gets charged 5 x .897 runs. So aren't you penalizing starters for having lousy relievers, or who were just unlucky that their relievers failed them? Wouldn't it be better to automatically just charge the starters those runs based on the chances that those runners he left were to actually score? This way, it isn't dependent on the quality of the relief performance.

It's a matter of interpretation; but interpretation is critical here.

When the starter leaves the game in the exact situation cited, he is inked in for 5 runs scored against him. He is also penciled in for the .897, because that is the effective scoring potential he surrendered to the opposition. Thus, the 5 = reality. On the 5.897, the jury is still out.

Now comes the reliever, arriving with +/0 .000 before he throws his first pitch. If he retires the side [2/3 of an IP] with no advance from third base, he inherited a .897 challenge; but the .897 is not a direct threat to the reliever's record. It is a virtual threat to the starter's record & the team's record.

If he retires the side, the situation is starter 5.0; reliever: -.897. Now a mathematical purist may well argue that this creates a mathematical inconsistency: 5-.897 = 4.103, which does not accurately reflect the score.

In this discussion, it would be fine with me to ink in the 5.897 agaist the starter because that's his actual game legacy. The reliever getting credit @ -.897 gets full credit for not letting a bad situation get worse.

That result would yield: starter +5.897; reliever, -.897 = actual score = 5.000. Neat! I prefer that. Thanks.

[Patriot]

Yes. The 89.7% is a percentage conversion of the decimal .897 cited for the base out situation, "B; 1 Out" in the Hidden Game of Baseball; p. 153.

No. The .897 value means that there is an average of .897 runs scored in the rest of the inning.

[leewileyfan]

No. The .897 value means that there is an average of .897 runs scored in the rest of the inning.

Thank you. I do stand corrected. I referred to the grid without re-reading on through the following page which clarifies the context.

The part that I had forgotten was that, with one out, there are two more outs to be recorded, so that the .897 is the sum of the runner on 3B PLUS the advancement potential of the next two batters.

To quote directly from The Hidden Game of baseball:

"This means that of the run value inherent in the situation "man on third, one out" (namely, .897), .249 resides with the batter(s) and .648 with the baserunner. In other words, a runner on third with one out will score, on average, 64.8% of the time." [pages 153-4].

Given that correction, I'd like to see the grid used, employing the .648 [or current equivalent value] for the starter legacy and the reliever inheritance.

Similarly, the isolated metric contents could be used for batter performance, before and after.

[Los Bravos]

Nobody argues that the stat is valueless, just of relatively weak value as compared to a wealth of other data that is just as easily and widely available.Actually, Keith Law recently compared the stat to a pitcher's shoe size as a measure of worth. I have used that same metaphor before, as sort of an extreme example. I always felt it was an exaggeration that wasn't likely to be used by anyone in reality so I was a little surprised to see him come up with it and use it seriously.

I'm considering suing him and if I win, in lieu of payment, I'll have him make me an omelette :cool:

http://espn.go.com/sportsnation/chat/_/id/29526/mlb-insider-keith-law

[curveball]

Rating & evaluating pitcher performance is always both interesting & challenging.
This response has to do with starting pitchers only, and then, only those whose careers lay dominantly within the 1901-2009 time span.

The numbers below attempt to display starters from two perspectives:

1. Legacy, Career, "Greatness Points" accumulated points, which attempts to take the relative adjusted ERA for each pitcher and apply that to a longevity factor. Therefore, guys with extended, effective careers get special notice IF they generally maintained a kind of dominance over their extended numbers of years.

2. ERA, from whatever era, adjusted to all eras, for an isolated one-on-one comparison to any/all other pitchers under consideration; like a barroom debate along the lines of "What if Pitcher N faced off against Pitcher Q, each on one of his best days?

This whole post was prompted by STL's mention of Jack Morris and Catfish Hunter, whom I get to at the end.

Pitcher.........................Era-Adjusted ERA..............Legacy Points

W. Johnson..................... 2.25.............................1,479.2
Clemens......................... 2.54..............................1,072.7
Maddux.......................... 2.55..............................1,062.3
Alexander....................... 2.81.............................. 976.1
Grove............................ 2.30............................. 964.1
Mathewson..................... 2.77.............................. 918.0
Seaver........................... 2.71.............................. 908.4
Hubbell........................... 2.34.............................. 861.4
*Johnson, R................... 2.78................................748.2*
P. Martinez...................... 2.12.............................. 746.2


*Randy Johnson's omission on original list now corrected.

*C. Young........................ 2.67............................. 740.2

*[Cy Young's numbers are reduced because all seasons prior to and through 1900 are not included.

Spahn............................ 3.25............................. 731.1
E. Walsh......................... 2.41............................. 686.9
Smoltz........................... 2.78............................ 650.3
Feller............................. 3.02............................. 630.1
Ryan.............................. 3.51............................. 594.4
Vance............................ 2.75............................. 577.4
Lyons............................. 3.02............................. 556.1
Schilling.......................... 3.02............................. 537.4
Ruffing............................ 3.40............................. 533.1
Mussina.......................... 3.16............................. 530.2
K. Brown......................... 3.03............................. 521.4
Derringer........................ 3.15.............................. 505.9
Joss.............................. 2.55............................. 504.0

IMHO, any starting pitcher accumulating >500 Legacy/Career "Points," should be considered a no-brainer inductee into the HoF. In fact, given the group that follows, I'd argue for any pitcher with >400 such "points" should also be in on sheer volume of accomplishment, maintained competitive superiority[regardless of career W-L record, and having longevity with such performance.

E. "W" Ford................... 3.08............................. 498.8
Glavine......................... 3.49.............................. 496.5
M. Brown...................... 3.12............................. 488.1
"Dutch" Leonard............. 3.14............................ 487.1
G. Perry........................ 3.70............................. 477.0
Gibson.......................... 3.40............................. 474.7
Warneke....................... 2.97............................. 472.1
Carlton......................... 3.69............................. 469.5
Palmer.......................... 3.45............................. 461.4
Roberts......................... 3.64............................. 446.3
Gomez.......................... 2.91.............................. 443.0
Wilhelm......................... 2.82.............................. 420.0
Jenkins......................... 3.64............................... 427.6
Blyleven....................... 3.73............................... 426.1
Walters......................... 3.28.............................. 421.2
*Eckersley..................... 3.37............................... 413.5

*[Eckersley, renowned as a reliever, is here because he was, for years, an effective starter ........... no reliever, isolated in that role, has >400 "points"]

Marichal........................ 3.46............................... 406.5
C. Mays........................ 3.30............................... 402.9

Many HoF are not on these lists; and several non-HoF are. I don't believe a well-considered argument can be made against any of these.

Now, as to Jack Morris and Jim "Catfish" Hunter. Neither rates > 300 or even > 200 "points in this evaluation metric. This does not condemn either or both as bad pitchers; but I believe it makes an honest evaluation of their performance against the standards set for the metric.

"Catfish" Hunter............. 4.19.............................. 119.0
Jack Morris................... 4.16............................. 143.9

For context, here are other starters within this range:

Blackwell [108.24; 3.76]; Hershiser [124.0; 4.14]; Bender [159.9; 4.02]; Koosman [124.7; 4.21]; Trucks [166.31; 3.94]; Chance [151.4; 3.87]; Shantz [126.2; 3.92]; P. Niekro [146.1; 4.26]; John [122.4; 4.27]; C. Finley [118.0; 4.13].

Sometimes, excellent pitchers have gotten buried in smaller media franchise cities; have ground out workhorse efficiencies over long careers with feckless clubs. I believe this evaluation method shines the light on some of the most deserving forgotten men.


Just curious as to how Eckersley can rate so high. 413.5 points for a starter who threw approximately 2500 innings at an era+ of 111 seems out of line. David Cone threw close 2900 innings at an era+ of 120.

In my opinion, Cone was a much better starter than Eckersley. How many legacy points does Cone have.

Also, I believe some pitchers are unfairly being punished for pitching longer. Blyleven's best years were the 70s. How many legacy points would he have if you only accounted for his pitching in the 70s? He pitched 2624 innings at an era+ of 130, and that is much better than Eckersley's 2500 innings at an era+ of 111.

Dave Stieb threw 2900 innings at an era+ of 122, I would have also expected him to have more legacy points than Eckersley.

[leewileyfan]

Just curious as to how Eckersley can rate so high. 413.5 points for a starter who threw approximately 2500 innings at an era+ of 111 seems out of line. David Cone threw close 2900 innings at an era+ of 120.

In my initial listing, I deliberately included this footnote:


*[Eckersley, renowned as a reliever, is here because he was, for years, an effective starter ........... no reliever, isolated in that role, has >400 "points"]

I had considered leaving out any pitcher, especially any noted for relief work; but Eckersley's high overall rating prohibited me from knowingly omitting him.

His rating includes his entire body of work, starter + reliever. Some may debate this hotly; but I have attempted to construct an evaluation system that numerically segregates starters from relievers. With guys like Wilhelm, Fingers and Eckersley, overlap is hard to avoid.


In my opinion, Cone was a much better starter than Eckersley. How many legacy points does Cone have.

I'm a David Cone fan. I have him @ 269.05 legacy points with a career ERA adjusted to 3.34.

Dave Stieb is @ 240.14, with ERA @ 3.22.

For across-the generations similarities, here are pitchers in the same neighborhood, pretty good company, I'd say:

Messersmith: 250.08; 3.07
Guidry: 261.32; 3.04
Tiant: 256.57; 3.22
Chandler: 273.83; 2.85
Hughson: 252.77; 2.82
Appier: 275.78; 3.05

You may note that Chandler & Hughson had abbreviated careers. There are some who would argue against their appearance on any list for relative IP. I am not among that school. I take the pitcher's individual career legacy; make adjustements for hitting-pitching climates; and let the numbers speak for themselves.

Example, Chandler came up at an advanced age and then lost 1.5 years to military service in WW II.

Hughson was a bona fide ace, who served 1 year in WW II and came back with an injured arm. They need no excuses, like lack of a decline period.


Also, I believe some pitchers are unfairly being punished for pitching longer. Blyleven's best years were the 70s. How many legacy points would he have if you only accounted for his pitching in the 70s? He pitched 2624 innings at an era+ of 130, and that is much better than Eckersley's 2500 innings at an era+ of 111.

I am a pro-Blyleven/HoF guy; but a pitcher's career legacy is largely based on decisions he makes with the MLB marketplace at the time. If a guy stays on beyond his effectiveness, that's part of his legacy.

Eckersley's overall rating included his impressive relief IP.


Dave Stieb threw 2900 innings at an era+ of 122, I would have also expected him to have more legacy points than Eckersley.

Same response applies.

P.S.

You relate all your responses to ERA+; so, here's how all pitchers mentioned fared [career] in that department:

Stieb: 122
Cone: 1.20
Messersmith: 121
Guidry: 119
Tiant: 114 [hung around 3 years too long]
Chandler: 132
Hughson: 125
Appier: 121

As for Eckersley, I'll take your word for 111 as a starter. However, career, he was 116. His relief work [less volume of IP related to the whole was E: 130-140.

[leewileyfan]

In drafting my reply [above] to curveball, I came to realize that I had totally missed a picther from the past, another who had dominant [starter] and later [relief] career portions.

The pitcher, one whom my father often praised, was the Pride of Havana,
Dolf Luque. I just ran this mighty mite's [5'7"] numbers and was astonished with both the legacy points and the generation-adjusted ERA.

Luque comes in at 521.46; 3.04, for me a no-brainer 1st ballot HoF-er. I was shocked; but his career numbers do support a case for enhanced reputation.

Have added him to my post-listing earlier in thread.

ERA+ through age 44 [stayed too long] = 117.

My nearest comp: Kevin Brown.

[curveball]

In my initial listing, I deliberately included this footnote:



I had considered leaving out any pitcher, especially any noted for relief work; but Eckersley's high overall rating prohibited me from knowingly omitting him.

His rating includes his entire body of work, starter + reliever. Some may debate this hotly; but I have attempted to construct an evaluation system that numerically segregates starters from relievers. With guys like Wilhelm, Fingers and Eckersley, overlap is hard to avoid.



I'm a David Cone fan. I have him @ 269.05 legacy points with a career ERA adjusted to 3.34.

Dave Stieb is @ 240.14, with ERA @ 3.22.

For across-the generations similarities, here are pitchers in the same neighborhood, pretty good company, I'd say:

Messersmith: 250.08; 3.07
Guidry: 261.32; 3.04
Tiant: 256.57; 3.22
Chandler: 273.83; 2.85
Hughson: 252.77; 2.82
Appier: 275.78; 3.05

You may note that Chandler & Hughson had abbreviated careers. There are some who would argue against their appearance on any list for relative IP. I am not among that school. I take the pitcher's individual career legacy; make adjustements for hitting-pitching climates; and let the numbers speak for themselves.

Example, Chandler came up at an advanced age and then lost 1.5 years to military service in WW II.

Hughson was a bona fide ace, who served 1 year in WW II and came back with an injured arm. They need no excuses, like lack of a decline period.



I am a pro-Blyleven/HoF guy; but a pitcher's career legacy is largely based on decisions he makes with the MLB marketplace at the time. If a guy stays on beyond his effectiveness, that's part of his legacy.

Eckersley's overall rating included his impressive relief IP.




Same response applies.

P.S.

You relate all your responses to ERA+; so, here's how all pitchers mentioned fared [career] in that department:

Stieb: 122
Cone: 1.20
Messersmith: 121
Guidry: 119
Tiant: 114 [hung around 3 years too long]
Chandler: 132
Hughson: 125
Appier: 121

As for Eckersley, I'll take your word for 111 as a starter. However, career, he was 116. His relief work [less volume of IP related to the whole was E: 130-140.


My fault, I thought that Eckersley's total was based strictly on his starting career, and no relief considerations.

It is just my thinking that Blyleven may have had a higher legacy point total if his career had ended in the 70s, and he never threw a pitch in the 80s. If that is the case, then it seems that he is being punished for pitching longer, even though he was still decent enough in the 80s. This not only applies to Blyleven, but other pitchers who had long careers. Phil Niekro gets absolutely crushed in your system.

So my question really is, let us say that,

2 pitchers have identical careers in their first 10 years

Same era+ of 150, same whip, etc...

So lets say pitcher A has 400 legacy points
and pitcher B has 400 legacy points

Pitcher A retires, so his legacy points remains at 400
Pitcher B goes on to pitch another 10 years, but during that period his era+ was only 110. Would his legacy points increase or decrease from the 400 he had?

Which is really why I asked if Blyleven's legacy points for his pitching in the 70s only would have been higher than the career total of 426.1 that you have given him.


Baseball-reference has a very nice yearly cumulative era+ which allows you to see starters era+ change as their careers evolve. You have to click on the "more stats" option on the player's page.

[leewileyfan]

My fault, I thought that Eckersley's total was based strictly on his starting career, and no relief considerations.

It is just my thinking that Blyleven may have had a higher legacy point total if his career had ended in the 70s, and he never threw a pitch in the 80s. If that is the case, then it seems that he is being punished for pitching longer, even though he was still decent enough in the 80s. This not only applies to Blyleven, but other pitchers who had long careers. Phil Niekro gets absolutely crushed in your system.

Since my evaluations are primarily focused on ERA equalization across several disparate generations of hitting climates, some pitchers may seem to suffer in that adjusted ERA's may look somewhat removed from ERA's in the season-by-season logs.

Secondly, legacy points are weighted largely by game equivalents of IP, career. They are also weighted to accord a split between starters and relievers, whom I attempt to segregate except for those who had large career chunks in both categories.

The longevity issue assures that shortened-career guys are not propelled like rockets into the top tiers; but I believe this equalizer has been applied equitably, so they are not cheated, either.


So my question really is, let us say that,

2 pitchers have identical careers in their first 10 years

Same era+ of 150, same whip, etc...

So lets say pitcher A has 400 legacy points
and pitcher B has 400 legacy points

Pitcher A retires, so his legacy points remains at 400
Pitcher B goes on to pitch another 10 years, but during that period his era+ was only 110. Would his legacy points increase or decrease from the 400 he had?

Which is really why I asked if Blyleven's legacy points for his pitching in the 70s only would have been higher than the career total of 426.1 that you have given him.

Rather than take two such 400 models, I thought it would be more revealing to look at actual pitchers, whose dynamics hover @ +/- 400.

Drysdale: Benefiting from weak-hitting climate pf '64-'68, he remained essentially consistent through better hitting climates through a career of > 3,400 IP. Only his last season, @ 5-4 would tend to be a slight drag on his overall number, while the earlier weak-hitting climate would affect his number a bit moreso. Final comment: Drysdale @ 378; was probably at or near that level steadily, not a 400+ pitcher pulled down by some extended period of failure.

Marichal: Marichal @ 406 may have been pulled down from 415 or so by three seasons too long on his record.

Koufax: Awful and erratic start followed by awesome dominance, somewhat diluted by awful batting in mid through late '60s. Koufax is the template for career dynamics being a two-way street, Sudden, tragic career ending due to arthritis, caught a star rising fast into a heightened offensive climate just at the moment he had a solid shot at moving over 400+ .... maybe more.

Mordecai Brown: Magnificent 1904-1911 stretch, diluted by '07 and '08 [Koufax, 60 years earlier]. His 488 could have been 500+; but there is slight dilution from that weak hitting span.

Rube Waddell: Injury seasons 1912 & 1916, done at age 32 = Mordecai Brown - innings pitched longevity = 387. [See also, Koufax].

Blyleven: a workhorse with internal pitching dynamics generally as steady as a gyroscope. Two seasons in prime years lost largely to injury. His 426 looks about as steady a rating as any pitcher listed.

Jenkins: '67 thru '69, weak hitting climate dilutes dominance; but steadiness thereafter shoots him to the 440-450 class, with last three season tail-off ending his record @ 428. If he's in, Blyleven should be, too.

Bunning: A 400 rating diluted to 368 by four season tail-off at end of career.

Early Wynn: [See Blyleven], a workhorse @ 400 + through a long career diluted by all those years trying to get #300 ........ finally.

Dizzy Dean: A short-career rocket ending up @ 378, with career interrupted in prime by All Star game foot injury and struggles discounting earlier excellence.

IMO, the numbers make the legacy and vice versa. It's like an elevator, some go up; some go down. Some is self-willed extension; some is successful longevity; and some is tragic interruption.


Baseball-reference has a very nice yearly cumulative era+ which allows you to see starters era+ change as their careers evolve. You have to click on the "more stats" option on the player's page.

Yep. B-R is one of my favorite sites.

[digglahhh]

Actually, Keith Law recently compared the stat to a pitcher's shoe size as a measure of worth. I have used that same metaphor before, as sort of an extreme example. I always felt it was an exaggeration that wasn't likely to be used by anyone in reality so I was a little surprised to see him come up with it and use it seriously.

I'm considering suing him and if I win, in lieu of payment, I'll have him make me an omelette :cool:

http://espn.go.com/sportsnation/chat/_/id/29526/mlb-insider-keith-law

Keith always stirring the pot, huh. As is clear from the discussion, I agree in sentiment. Though, if pressed to articulate my views in legalese, so to speak, I drop the hyperbole and speak very deliberately about my opinions regarding wins. It doesn't make for very fun reading.

I'd use the shoe size line too... in a safe room.

My girlfriend just met Keith Law at the Winter Meetings actually. She spoke with both Law and Will Carrol spoke about CYA-Gate.