Basic Adjusted ERA and OPS Question

[Kitty Kaat]

If an Adjusted ERA is 130 does that mean the pitcher is 30% better than the average pitcher at that time? Same with Adjusted OPS, if David Wright is at 140 is he 40% better than the average hitter?

[Brad Harris]

If an Adjusted ERA is 130 does that mean the pitcher is 30% better than the average pitcher at that time? Same with Adjusted OPS, if David Wright is at 140 is he 40% better than the average hitter?
Using your examples, an Adjusted ERA of 130 indicates the pitchers' ERA is 30% better (i.e. lower) than the league average. The OPS+ of 140 demonstrates that Wright's OPS is 40% better (i.e. higher) than league average. Making the distinction between that stat and the player, but yes...looks like you've got the gist of it.

[mwiggins]

Using your examples, an Adjusted ERA of 130 indicates the pitchers' ERA is 30% better (i.e. lower) than the league average. The OPS+ of 140 demonstrates that Wright's OPS is 40% better (i.e. higher) than league average. Making the distinction between that stat and the player, but yes...looks like you've got the gist of it.

Actually, it's not exactly his OPS that's 40% better. His SLG and OBP are, combined, 40% better than the league average.

For example, David Wright's current OPS+ is 143, but his OPS is only 21% better than the park adjusted league average. His SLG is 16% better than the league, and his OBP is 27% better - which gets you the OPS+ of 143.

[brett]

Using your examples, an Adjusted ERA of 130 indicates the pitchers' ERA is 30% better (i.e. lower) than the league average. The OPS+ of 140 demonstrates that Wright's OPS is 40% better (i.e. higher) than league average. Making the distinction between that stat and the player, but yes...looks like you've got the gist of it.

Also, in addition to what's been written, an OPS+ of 140 happens to correlate fairly well to a production level 40% above league average.

The correlation hold well for a) teams and b) OPS+ scores close to 100 which would be true of most stats based on an average of 100.

It doesn't work very well when you get real low either. A player can have a 0 OPS+ by slugging half the league average and getting on base at half the league average, but he's still going to produce some runs.

As far as ERA+, 130 doesn't really mean 30% better.

What it means is that he gives up 1/1.3 as many earned runs as average which is an ERA that is 76.9% of the league average, or 23.1% lower than average.

Take another example, a guy with a 150 ERA+ and a guy with a 200 ERA+.

The guy at 150 is giving up 1/1.5 or .67 league average and the guy at 200 is giving up .50 the league average.

[Tango Tiger]

ERA+ of 130 means that the LEAGUE gives up 30% more runs than the pitcher.

Guess what an ERA+ of 200 means? Right, the league gives up 100% more runs than the pitcher (i.e., the pitcher givs up runs at half the league rate).

And 300 means the league gives up 200% more runs, or the pitcher gives up runs at one-third the league rate.

Confused? If you don't like it, complain about it.

[SABR Matt]

Both OPS+ and ERA+ are poorly designed for various different reasons that have been discussed at length here. They made decent quick-look thumbnail metrics but they can mislead you if you aren't careful.

[Second Base Coach]

Both OPS+ and ERA+ are poorly designed for various different reasons that have been discussed at length here. They made decent quick-look thumbnail metrics but they can mislead you if you aren't careful.

Poorly designed? Whatsyerbeef with ERA+?

[brett]

Poorly designed? Whatsyerbeef with ERA+?

I would prefer relative ERA, or relative ERA on a 4.00 average scale.

Take a guy with a 100 ERA+ one year and a 200 ERA+ the next.

Then another guy with 2 years both of 150.

The first guy actually will have a total ERA+ of 133 while the second guy will be at 150.

Player 1:
year 1: 1.00 relative era=100 ERA+
year 2: 0.50 relative era=200 ERA+
avg yr: 0.75 relative era=133 ERA+

Player 2:
year 1: 0.67 relative era=150 ERA+
year 2: 0.67 relative era=150 ERA+
avg yr: 0.67 relative era=150 ERA+

Another way to look at it is that the 150 ERA+ season is saving 33% of the league average runs per game, while the 200 season is saving 50%. The 200 season looks twice as much better than the league, but is only saving "half again" as many runs, not twice as many.

Of course the other issue is that ERA+ may deviate by up to 10% from a pitchers true worth over his career dependent on defense.

And probably by 20% over a single 200 IP season.

[SABR Matt]

I wasn't talking about defensive context or the improper use of park-run factors, but of which are MAJOR sources of bias for ERA+ even at the career level. I was talking about essentially the same thing you're talking about...the need to switch ERA+ to a relative ERA scale to make it easier for people to equate a player to average in a framework that is accurately linear.

[JRB]

Both OPS+ and ERA+ are poorly designed for various different reasons that have been discussed at length here. They made decent quick-look thumbnail metrics but they can mislead you if you aren't careful.

That's interesting Matt. I've come to regard OPS+ as an outstanding metric, though no metric is without its flaws. It does seem to favor hitters that draw a lot of walks. What batting metric do you think better evaluates hitters? And why?

I tend to think ERA+ to be not quite as reliable an indicator as OPS+, especially when comparing starting pitchers from the past. The reason is that the main idea for a starting pitcher was always to get a W. Some pitchers learned how to coast, and give a up a few unimportant runs here and there when they had big lead, or to make sure they completed the game, etc, but knew how to bear down and get the outs it takes to win. There are a number of outstanding pitchers from the past who were real winners, who don't look so hot when you just look at their ERA+. At the end of the day you probably want to have your money riding on the accomplished winner as opposed to somebody that just has a better looking ERA+ metric. I think with that stat there should be some correlation with how the pitcher actually was at winning games.

[STLCards2]

That's interesting Matt. I've come to regard OPS+ as an outstanding metric, though no metric is without its flaws. It does seem to favor hitters that draw a lot of walks. What batting metric do you think better evaluates hitters? And why?

I tend to think ERA+ to be not quite as reliable an indicator as OPS+, especially when comparing starting pitchers from the past. The reason is that the main idea for a starting pitcher was always to get a W. Some pitchers learned how to coast, and give a up a few unimportant runs here and there when they had big lead, or to make sure they completed the game, etc, but knew how to bear down and get the outs it takes to win. There are a number of outstanding pitchers from the past who were real winners, who don't look so hot when you just look at their ERA+. At the end of the day you probably want to have your money riding on the accomplished winner as opposed to somebody that just has a better looking ERA+ metric. I think with that stat there should be some correlation with how the pitcher actually was at winning games.

The biggest problems with OPS+ are: it ignores baserunning, overvalues SL% compared to OB%, and has shaky at best park factors. Is it better than BA or RBI's? Heck yes. Is it a good starting point to look at how good of a hitter a guy is overall? Yes. Is it servicable for most fans when comparing or ranking players? Yes. Is it as reliable as newer metrics that fix the above problems? No.

The biggest problems with ERA+ are: shaky park factors (again), it doesn't include how much the pitcher's defense affected his run prevention, doesn't account for BABIP luck, and doesn't account for the level of difficulty faced by the pitcher. And most importantly, without a BF or IP component, it doesn't tell you too much. Is it better than wins or ERA by itself? Heck yes. Is it a great starting point to look at how good of a pitcher a guy is overall? Yes. Is it servicable for most fans when comparing or ranking players? Yes. Is it as reliable as newer metrics that fix the above problems? No.

And there is a positive correlation between a good ERA+ and a good winning %...something like .7 or so (need fact check here). The problem is, most people think the correlation is much, much closer to 1.

[Tango Tiger]

The biggest problems with OPS+ are: it ignores baserunning,

Can't you also say that the biggest problem with OBP is that the HR and walk are valued at "1" and baserunning is also ignored? SLG excludes walks. This is not a bug, it is a design feature.

Similarly, OPS+ is DESIGNED to ONLY consider OBP and SLG. You can't then say it has a problem because SB and CS aren't in there!

The problem with OPS+ is in its weighting of OBP and SLG. Clearly, it is attempting to put the two together to get at hitting value, and it's not doing a good enough job.

[538280]

That's interesting Matt. I've come to regard OPS+ as an outstanding metric, though no metric is without its flaws. It does seem to favor hitters that draw a lot of walks. What batting metric do you think better evaluates hitters? And why?



OPS+ does not favor batters who draw walks at all. OPS underrates the walk in comparison to the single because it essentially counts a walk as 1 (once in OBP) and a single as 2 (once in OBP and once in SLG), so a walk is worth 50% of a single. Linear Weights have shown that a walk is really worth about 70-72% of a single. OPS+ does fix this somewhat by essentially putting OBP on the same scale as SLG by adding the two relative rates together, but it probably still underrates the walk somewhat just like plain OPS does.

To me the best hitting metric is something like Linear Weights or BaseRuns, that takes the correct run values of each offensive event contributed by the player. Linear Weight runs above average are displayed on BBRef as BtRuns. This doesn't include baserunning, but you can figure basestealing runs pretty easily by taking (as an estimate, the real values probably vary by era) (.22*SB)-(.38#CS). It also of course doesn't factor in non-SB baserunning which can also be a factor.

EDIT: (and BTW I don't know how this heading got on the top, it may have been a mistake or something but please don't take it as an insult or anything)

[STLCards2]

Similarly, OPS+ is DESIGNED to ONLY consider OBP and SLG. You can't then say it has a problem because SB and CS aren't in there!



Yes...it is designed to only consider OBP and SLG, so it shows exactly what it is intended for. But you know that when we say that it is "flawed", that we mean specifically that it is being used my the majority or interpreted by the majority incorrectly. If somebody tries to use a tape measurer to hammer a nail, it is the workers fault for misuse, but that tape measurer is still a flawed tool for that specific job because of its specific qualities. The original poster did not ask how good OPS+ does at expressing OB% and slugg%, but how it it did at evaluating hitters productivity. One of the ways it falls short of his goal, is that it doesn't show baserunning. If he asked me if OB% and slugg% are a great tool for showing how good a guy's combined OB% and slugg% are, I would have said, "its the best!"

Also, I did say that OPS+ weighs OB% and Slug% incorrectly in my post.

[JRB]

OPS+ does not favor batters who draw walks at all. OPS underrates the walk in comparison to the single because it essentially counts a walk as 1 (once in OBP) and a single as 2 (once in OBP and once in SLG), so a walk is worth 50% of a single. Linear Weights have shown that a walk is really worth about 70-72% of a single. OPS+ does fix this somewhat by essentially putting OBP on the same scale as SLG by adding the two relative rates together, but it probably still underrates the walk somewhat just like plain OPS does.

To me the best hitting metric is something like Linear Weights or BaseRuns, that takes the correct run values of each offensive event contributed by the player. Linear Weight runs above average are displayed on BBRef as BtRuns. This doesn't include baserunning, but you can figure basestealing runs pretty easily by taking (as an estimate, the real values probably vary by era) (.22*SB)-(.38#CS). It also of course doesn't factor in non-SB baserunning which can also be a factor.

EDIT: (and BTW I don't know how this heading got on the top, it may have been a mistake or something but please don't take it as an insult or anything)

Is this Chris? Great to hear from you. Hope everything has been going well for you. I just recently started posting again on this site after a long hiatus (The usual culprit of work and women).

As far as the value of walks are concerned, this seems distinctly deja vu. We've had diiscussions on this topic before a long time ago. As I recall there is a lot of disagreement on what the true value of a walk is, perhaps it is even less than half the value of hit, and since hits include extra base hits, maybe much less than half the value. The walk is not counted in SA, but neither is the AB. Isn't SA really measuring the quality of a player's hits? Why should walks be any part of that? For really dangerous hitters like Ruth, Williams, Mantle, Bonds, etc who are being walked because they are being pitched around they could make the argument that by being walked they are being deprived of quite a few extra base hits because the pitchers are avoiding pitching to them, though perhaps this argument is not valid unless you can show that they would have hit extra base hits at a higher rate if they had been pitched to. Also, there are a number of hitters who are being walked for other reasons such as they have an exceptional batting eye, or have small or odd size strike zone. In the case of these hitters getting a walk is probably the best they can hope to achieve, so that argument wouldn't work for them.

I've aware of the batting runs stat. I like using the Official Encyclopedia and they show batting runs for every player. I notice that the batting runs shown on the Baseball Reference website differ from the batting run figures found in the Official Encyclopedia. I guess this is similar to the discrepancy between the OPS+ figures shown in the Official Encyclopedia and the OPS+ figures shown on Baseball Reference. With regard to OPS+ I notice that the the guy from baseball reference comments that he has devised a much more complicated formula than that used by the Baseball Encycopedia, so I surmize that he may once again have created his own home brew for his batting runs formula. In the past I came across a number of instances where his home grown OPS+ calculations didn't seem to pass the smell test so I'd be a bit cautious about using his batting run calculaitons instead of those in the Official Encyclopedia.

Chris. Regardless of whether we take the version of the batting runs metric from the Official Encyclopedia or the guy over at Baseball Reference, what specifically is it about the batting run metric that you feel makes it superior to OPS+.

I also notice that the Official Encyclopedia has another stat that combines batting runs, fielding runs, and baserunning. I don't have the book right in front me but I believe its called BRFR or something along those lines. What do you think of that metric for rating a player's overall game. I do notice that some of the players you used to (and perhaps still do) think quite hiighly of fair very well with this metric. I'd welcome hearing your thoughts on the subject.

[538280]

Is this Chris? Great to hear from you. Hope everything has been going well for you. I just recently started posting again on this site after a long hiatus (The usual culprit of work and women).

It is Chris. I also took an extended leave but now am back.


As far as the value of walks are concerned, this seems distinctly deja vu. We've had diiscussions on this topic before a long time ago. As I recall there is a lot of disagreement on what the true value of a walk is, perhaps it is even less than half the value of hit, and since hits include extra base hits, maybe much less than half the value. The walk is not counted in SA, but neither is the AB. Isn't SA really measuring the quality of a player's hits? Why should walks be any part of that?

There is no disagreement on what the value of a walk is vs. a single, double, triple, HR, etc. Linear Weights credit a walk as worth about .33 runs in a typical environment and a single as worth about .47. That means a walk is worth 70.21% of a single. I wasn't speaking in my post of the value of a walk vs. that of any hit, but that of a walk vs. a single. OPS counts a walk as one and a single as two, so it counts a walk as 50% of a single, less than the 70.21% of a single that a walk really should be counted as.

Yes, SLG is meant to measure the quality of hits and thus it on its own should not include walks. SLG is not meant to be an evaluation of a hitter's total contribution, so that it has exactly proper run values isn't important. However OPS is used is as an evaluation of a hitter's total contribution to his team, so if it is to be used its run values should be accurate, and in the case of its value to of a single vs. a walk is overvalues a single in comparison with a walk.

For really dangerous hitters like Ruth, Williams, Mantle, Bonds, etc who are being walked because they are being pitched around they could make the argument that by being walked they are being deprived of quite a few extra base hits because the pitchers are avoiding pitching to them, though perhaps this argument is not valid unless you can show that they would have hit extra base hits at a higher rate if they had been pitched to. Also, there are a number of hitters who are being walked for other reasons such as they have an exceptional batting eye, or have small or odd size strike zone. In the case of these hitters getting a walk is probably the best they can hope to achieve, so that argument wouldn't work for them.


I've aware of the batting runs stat. I like using the Official Encyclopedia and they show batting runs for every player. I notice that the batting runs shown on the Baseball Reference website differ from the batting run figures found in the Official Encyclopedia. I guess this is similar to the discrepancy between the OPS+ figures shown in the Official Encyclopedia and the OPS+ figures shown on Baseball Reference. With regard to OPS+ I notice that the the guy from baseball reference comments that he has devised a much more complicated formula than that used by the Baseball Encycopedia, so I surmize that he may once again have created his own home brew for his batting runs formula. In the past I came across a number of instances where his home grown OPS+ calculations didn't seem to pass the smell test so I'd be a bit cautious about using his batting run calculaitons instead of those in the Official Encyclopedia.

The way BBRef does it is actually more correct IMO. What the Encyclopedia does is apply the batting park factors to the league OBP and SLG. This isn't really correct because those park factors are factors for runs-not OBP and SLG. OBP and SLG by themselves aren't necessarily affected by the parks the same way that runs are. What BBRef has done is essentially applied the park factor to runs created, and then when it is applied see how the league OBP and SLG change. This is more correct because it is applying the park factor to runs rather than OBP and SLG. The park factor is meant to be applied to runs, applying it to OBP and SLG as the Encyclopedia does isn't really correct.


Chris. Regardless of whether we take the version of the batting runs metric from the Official Encyclopedia or the guy over at Baseball Reference, what specifically is it about the batting run metric that you feel makes it superior to OPS+.

It is Linear Weights, meaning it takes the actual accurate run value of each event, as studied through PBP data, and applies those values to each of the batter's own offensive events. It has values for each offensive event that are truly accurate and have been backed up through real research of the game, not something which just works well for a fast and quick evaluation like OPS or OPS+.


I also notice that the Official Encyclopedia has another stat that combines batting runs, fielding runs, and baserunning. I don't have the book right in front me but I believe its called BRFR or something along those lines. What do you think of that metric for rating a player's overall game. I do notice that some of the players you used to (and perhaps still do) think quite hiighly of fair very well with this metric. I'd welcome hearing your thoughts on the subject.

I think it is okay but there are a number of problems with it due to its defensive system (Fielding Runs) which doens't really make any sense, it doesn't adjust for any of the defensive contexts that more advanced defensive metrics do take into account, it kind of just uses a range-based defensive analysis based on plays made per inning (and often uses bad estimates for innings when the data isn't available). It also bases the system on average which isn't really the greatest either. You have to just realize going in that it is the number of games above or below average for a career. The problem is that some players spend much time particularly at the end of their career where they are average or slightly below average and yet get no credit on their career total. They aren't great anymore, but they are still making a very important contribution to their teams, yet they will either have no change or come down in that system because it sets its baseline at average. So, overall, I don't agree with setting the baseline at average.

[SABR Matt]

Hello Chris...haven't seen you around in a LONG time.

You and I have had some good conversations in the past...good to see you posting again.

You are spot on as to the problems with TPR (the encyclopedia lists TPR) and with OPS+. I think BRef is still doing it wrong when it comes to park-adjusting OPS+...using a ratio-factor designed to adjust runs to make a new projection for OBP and SLG seems fraught with peril. The better approach, IMHO would be to convert OPS into batting runs (or just use BtRuns...LOL), apply the park factor to the league average batting run rate, and take your ratio. Extrapolating a run-factor to a new league OBP and SLG is mixing apples and oranges.

Incidentally, I've found that extreme parks have pretty bad OPS+ biases relative to PCA win rates (hitters in hitting parks get badly overestimated by OPS+ compared to a total player run created metric ratio'ed to league average production rate and the reverse is true for pitcher's parks). I think this is happening because something is going wrong with how they adjust OBP and SLG expectations...a BtRuns to park adjusted average BtRuns ratio has a better chance of being correct since there is no conversion going on.

[JRB]

It is Chris. I also took an extended leave but now am back.



There is no disagreement on what the value of a walk is vs. a single, double, triple, HR, etc. Linear Weights credit a walk as worth about .33 runs in a typical environment and a single as worth about .47. That means a walk is worth 70.21% of a single. I wasn't speaking in my post of the value of a walk vs. that of any hit, but that of a walk vs. a single. OPS counts a walk as one and a single as two, so it counts a walk as 50% of a single, less than the 70.21% of a single that a walk really should be counted as.

Yes, SLG is meant to measure the quality of hits and thus it on its own should not include walks. SLG is not meant to be an evaluation of a hitter's total contribution, so that it has exactly proper run values isn't important. However OPS is used is as an evaluation of a hitter's total contribution to his team, so if it is to be used its run values should be accurate, and in the case of its value to of a single vs. a walk is overvalues a single in comparison with a walk.




Chris: I know that you feel that somehow you're getting shortchanged in OPS+ by the fact that walks and singles are taken into account in the OBP component and only singles are taken into account in the SA component. However, I don't think that's really the case because SA is a rate stat not a cumulative stat. And just because they add SA and OBP together to get OPS doesn't make OPS a cumulative stat. All you are doing is adding together two rate stats Adding together two rate states does not transform the result into a cumulative stat. It's merely a convenient way of using one number to relay the information that you would formerly have gotten by looking at two numbers. OBP tells you how good a player was at getting on base, and SA tells you how productive a hitter is when he swings the bat.

As I recall, you're a Rickey Henderson admirer, so let's use him as an example. Rickey has a superb career OBP of .401, added greatly by the fact he got so many walks. So all his walks are iincluded in the OBP component of OPS+!. Then we look at his SA of .419 which is good for a lead off hitter, though not nearly as impressive as his OBP. All of Rickey's hits are included in his SA. The fact that Rickey's walks are not included does not diminish his SA in any way, because SA is only a rate stat not a cumulative stat, He is not being treated differently than any other batter. The big difference is that in comparison to some hitters who walked a lot such as Ted Williams, Mickey Mantle, etc Rickey was not nearly as productive with his hits. Hitters like Williams and Mantle, who are both good at drawing walks and are also very productive when they hit the ball, come out looking great in the OPS+ metric. It's when a player is not quite as good in either drawing walks (classic example George Sisler) or being productive with his hits that his flaw shows up in the overall OPS+.

[JRB]

Hello Chris...haven't seen you around in a LONG time.

You and I have had some good conversations in the past...good to see you posting again.

You are spot on as to the problems with TPR (the encyclopedia lists TPR) and with OPS+. I think BRef is still doing it wrong when it comes to park-adjusting OPS+...using a ratio-factor designed to adjust runs to make a new projection for OBP and SLG seems fraught with peril. The better approach, IMHO would be to convert OPS into batting runs (or just use BtRuns...LOL), apply the park factor to the league average batting run rate, and take your ratio. Extrapolating a run-factor to a new league OBP and SLG is mixing apples and oranges.

Incidentally, I've found that extreme parks have pretty bad OPS+ biases relative to PCA win rates (hitters in hitting parks get badly overestimated by OPS+ compared to a total player run created metric ratio'ed to league average production rate and the reverse is true for pitcher's parks). I think this is happening because something is going wrong with how they adjust OBP and SLG expectations...a BtRuns to park adjusted average BtRuns ratio has a better chance of being correct since there is no conversion going on.

Hi Matt:

Actually the Encyclopedia no longer uses TPR (Total player rating). The newer 4th edition published in 2007 now uses BFW which they define as follows

"the sum of a player's batting wins, basestealing wins, and fielding wins, this figures indicates how many games the player won or lost for his team compared to an average player."

In the older editions Total Player rating was defined as follows: "This is the sum of a player's adjusted batting runs, fielding runs, and base stealing runs, minus his positional adjustment, all divided by runs per win factor for that year-generally around 10, historically around the 9-11 range.

It appears that BFW has a different formula than TPR. I've noticed some dramatic changes in the numbers for some players when using BFW rather than TPR.

With your statistical background, perhaps you can take a look at the BFW metric and let us know your thoughts on what the pluses and minuses are of that metric, and whether you feel it is an improvement on TPR..

[SABR Matt]

The critical component would be the fielding side...because TPR was abysmal with fielding. Where can I find a full disclosure of the BFW formula?

[538280]

Chris: I know that you feel that somehow you're getting shortchanged in OPS+ by the fact that walks and singles are taken into account in the OBP component and only singles are taken into account in the SA component. However, I don't think that's really the case because SA is a rate stat not a cumulative stat. And just because they add SA and OBP together to get OPS doesn't make OPS a cumulative stat. All you are doing is adding together two rate stats Adding together two rate states does not transform the result into a cumulative stat. It's merely a convenient way of using one number to relay the information that you would formerly have gotten by looking at two numbers. OBP tells you how good a player was at getting on base, and SA tells you how productive a hitter is when he swings the bat.



Of course it's not a cumulative stat, it's a rate because there is a dividing factor (though it is different for OBP and SLG, PA and AB, there is still a dividing factor really and it's opportunities). However any rate stat adds things together and then divides them. The adding part for OPS essentially is total bases plus times on base. That is the part that I was referring to when I meant that OPS counts a walk as 1 and a single as 2. A walk is one time on base, zero total bases, so it is worth 1. A single is one time on base, one total base, so it is worth 2.


As I recall, you're a Rickey Henderson admirer, so let's use him as an example. Rickey has a superb career OBP of .401, added greatly by the fact he got so many walks. So all his walks are iincluded in the OBP component of OPS+!. Then we look at his SA of .419 which is good for a lead off hitter, though not nearly as impressive as his OBP. All of Rickey's hits are included in his SA. The fact that Rickey's walks are not included does not diminish his SA in any way, because SA is only a rate stat not a cumulative stat, He is not being treated differently than any other batter. The big difference is that in comparison to some hitters who walked a lot such as Ted Williams, Mickey Mantle, etc Rickey was not nearly as productive with his hits. Hitters like Williams and Mantle, who are both good at drawing walks and are also very productive when they hit the ball, come out looking great in the OPS+ metric. It's when a player is not quite as good in either drawing walks (classic example George Sisler) or being productive with his hits that his flaw shows up in the overall OPS+.

All that I'm saying is that the value of each walk is not being represented in OPS in an equal manner to its actual value on the field, as shown in Linear Weights. Every walk that Rickey draws is not being weighted quite as much as it should in OPS, every single that George Sisler gets is getting weighted a little more than it should. If you have one player who has 100 singles and 50 walks, and another player who has 75 singles and 75 walks OPS will show the 100 singles player to be better than the 75 singles player by a greater degree than he really is. Obviously the 100 singles player is still better.

[538280]

The critical component would be the fielding side...because TPR was abysmal with fielding. Where can I find a full disclosure of the BFW formula?

It's in the glossary of the ESPN Baseball Encyclopedia, which can be found here:

http://247baseball.com/web/wp-content/uploads/2009/04/15-ban-bb2008-bm1.pdf

Look at the entry for Fielding Runs.

I'm not entirely familiar with what the old TPR Fielding Runs formula was, but if this is an improvement over that I can't imagine how bad that was. It is what I was referring to above, it is based almost entirely on the "fielding rates", which are basically just the raw defensive numbers. The only things they attempt to adjust for is the difference in left and right handed pitching on the teams, and they do make somewhat of an attempt at making up a "DP opportunities". The fielding rates also have a number of problems, for instance the 1Bman one is almost based entirely on assists, the same one is used for P, 2B, 3B, and SS. It just isn't a very good system, shouldn't be trusted at all, there are so many better fielding metrics out there....