238 replies to this topic

[CA-ORIOLE]

I feel like this article could have been one sentence long instead of ten thousand words or whatever it was:

 

Use WPA instead of WAR if you want to know how a player's contributions impacted his team's wins and losses.

Well, that's Bill James. He's just as much a philosopher than he his a sabremtrician. Some people love it and some people say that's why everybody has passed him by. 

 

To your point, I think you can use WPA (and/or other clutch stats/metrics) and adjust WAR if you want to do along the lines of what James is proposing. I think you have to take the idea of tea wins out of it. Way too hard and way too many variables imo. 

 

Not sure if that answers Matt's question or not.

[CA-ORIOLE]

The other plan would be instead of using a standard factor, to use leverage instead. The only problem is, suppose you have a pitcher that was huge in the clutch but had terrible results overall (-.3 WAR). You'd want to make sure he has positive value. So, you'd have to use WAR on the one hand and then build a clutch metric (based on WPA perhaps) that took into account the difference between pythagorean wins and actual wins. 

This approach sounds about right. Honestly I'm not sure I'd be all that interest in it, but I think it's better than what James is doing. I could be wrong, but I thin both versions of WAR already use leveraged performance for pitchers with relief pitchers having more substantive/relative impact. 

[Matt_P]

This approach sounds about right. Honestly I'm not sure I'd be all that interest in it, but I think it's better than what James is doing. I could be wrong, but I thin both versions of WAR already use leveraged performance for pitchers with relief pitchers having more substantive/relative impact. 

 

fWAR does attempt to use leverage but their methodology for doing so is just stupid. Fortunately, they make it such a small factor that it doesn't matter. But fWAR says that a pitcher used in higher leverage situation deserves amplified (not necessarily positive) WAR values. This would say that players on teams that overachieve deserve higher WAR values. Basically, instead of using LI like Fangraphs does, you'd go with WPA.

 

I'm not sure I like the idea so much either. 

[BSLSteveBirrer]

I still haven't seen anything beyond anecdotes that makes me think that teams that are boom or bust offensively are any worse off than teams that are consistent, assuming a similar total runs scored number. 

 

Not entirely sure how I'd even go about researching that theory.  BB-ref Play Index and Retrosheets would be a start, but it'd be a lot of work.  Would have to determine the variance of RS for each team, and if high variance correlates at all to high winning percentage.  But you'd only want to look at teams with similar total RS and RA.

This would be easy to do. Very time consuming but not difficult.  Just look at the number of runs scored in each game vs wins. How many times did a team score 0 runs and win (well clearly that one is easy), how many times did they score 1 run and win, etc. Look at the total distribution of runs scored and how they fared in W and L for each of those amounts. Then compare that to other teams with similar total runs scored. 

 

That would tell you whether or not a boom or bust offense is better, worse, or no different from an offense that is more consistent with higher lows and lower peaks.

 

My engineering instinct/gut tells me that if you have a team that scores 5 runs every single game during a 162 game season that they will have more wins than a team that scores the same 810 total runs but whose games are split between 10, 3, and 2 runs evenly.

 

I am guessing that a W L split would look something like this.

Score 10 is essentially 100% - 54 W

Score 3 is about a 30% - 16 W

Score 2 is about a 20% - 10 W

 

So this team would garner around 80 W. And I think I have been generous, it may actually be worse than this. The team that scores 5 each and every game should win over 50% of the time. Only 4 teams in the AL averaged better than 5.00 runs/game. Without getting into a bit of math I think this team would come in around 85 wins (and I think thats on the low side).

 

Now that is a pretty extreme set of runs scored/game but I think you get the gist of it. What would be a bit less work but might be interesting would be to look at the teams that all scored between 4.5 and 4.6 runs/game. Check their run distribution and W/L for each of those. I suspect that the O's would have more games where they scored a bunch and more games where they scored a few runs compared to the others.

[Mackus]

That would tell you how one boom or bust team did one year. Not a very compelling statistical case.

With the BB-Ref play index tool you could get the league distributions and team-by-team pretty easily.

[BSLSteveBirrer]

That would tell you how one boom or bust team did one year. Not a very compelling statistical case.

With the BB-Ref play index tool you could get the league distributions and team-by-team pretty easily.

True but you could do it for the last 5 years for the O's. You could do this for as many teams and for as many years as you wanted to in order to make it statistically valid. I would be willing to wager the data would prove my simple conjecture.

 

But hey I am also going on the eye test and perception. Maybe the O's scoring distribution isn't as much feast or famine as I think.

[Matt_P]

That would tell you how one boom or bust team did one year. Not a very compelling statistical case.

With the BB-Ref play index tool you could get the league distributions and team-by-team pretty easily.

 

You want Retrosheet not BB-Ref. I did this once upon a time. Not sure if I still have the data. 

[Mackus]

Both are annoying because they give you a string rather than a integer total of runs scored and allowed each game. Can't just easily delimit the csv and then do whatever formula you want (average, variance, sum, etc.) of an entire runs scored column.

Retrosheet gives a string for how many runs scored each inning, like 100000300 for visitor and 02100300x for home.

BB-ref gives a string for result and score, like W 6-4.

[Matt_P]

Both are annoying because they give you a string rather than a integer total of runs scored and allowed each game. Can't just easily delimit the csv and then do whatever formula you want (average, variance, sum, etc.) of an entire runs scored column.

Retrosheet gives a string for how many runs scored each inning, like 100000300 for visitor and 02100300x for home.

BB-ref gives a string for result and score, like W 6-4.

 

Try fields 10-11 Visiting and home team score (unquoted).

[dude]

The Ringer: The Next Battle in the War Over WAR

 

The Ringer does some good work.

 

Rany (author) asks the right question (among many good points) and everybody runs off to talk about some technical aspect of WAR.

[CA-ORIOLE]

The Ringer does some good work.

 

Rany (author) asks the right question (among many good points) and everybody runs off to talk about some technical aspect of WAR.

The article is primarily about Bill James proposal to re-allocate individual player WAR using team wins and run differential, for which he provides a basic framework, methodology and examples. While the entire article may reference tertiary aspects and background of WAR, James proposal is the primary emphasis. If you scroll down far enough you'll find the author essentially agrees with what some of us have said here, having some of the same criticisms. I don't think anyone, much less "everyone" has "just thrown out technical aspects of WAR" in response to and/or for the purpose of "ignoring the Authors good points", just that we've been generally focused to the primary emphasis of the article. 

 

FYI, I do agree with you that the author is pretty sharp,  but perhaps a bit long winded (but I'm guess you would find that a positive). 

 

Do you have an input as to James approach here? Or perhaps something else that you'd like to contribute that we all missed? 

[Mackus]

I don't think James is saying to replace WAR with one scaled for team wins relative to pythag wins. Rather, you should do that when determining who should be the MVP if you are using WAR to decide that.

[CA-ORIOLE]

I don't think James is saying to replace WAR with one scaled for team wins relative to pythag wins. Rather, you should do that when determining who should be the MVP if you are using WAR to decide that.

Well, he didn't mention pyhag as I recall, but it was relationship/adjustment of individual player value (specifically WAR) based on actual team  wins and run differential (I get those are the 2 elements of pythag and James invented it etc). While Judge and Altuve (both potential MVPs) were the main example/focus (the day before the MVP voting), I didn't gather the approach should only be applied to potential MVP's. I assumed he was setting that up as more of a glaring example to support his theory.  

[MDtransplant757]

Is it wrong that I prefer using bWar over fWAR? 

[CA-ORIOLE]

Is it wrong that I prefer using bWar over fWAR? 

With the recent changes to UZr I doubt you'll see as much variation with position players going forward. Depends on whether you like ERA based rWAR or FIP based fWAR. I prefeer rWAR, but often look at both. 

[dude]

The article is primarily about Bill James proposal to re-allocate individual player WAR using team wins and run differential, for which he provides a basic framework, methodology and examples. While the entire article may reference tertiary aspects and background of WAR, James proposal is the primary emphasis.

 

....in the sub title.

 

Over the past five years, WAR has caught on with the baseball-watching public in a way that no other sabermetric stat has. But as the recent debate about the MVP race between Aaron Judge and José Altuve revealed, we can’t fully move forward with the stat until we agree on what it’s meant to do.

 

You think it's about Judge - Altuve, but it's not.  They are an example in the article, but my issue with WAR is the horrible mis-application of the stat and what it actually means.

 

I get metrics.  Someday maybe I'll share.  I have a 40M plan to address challenges that have never been addressed in a 30B (actually much bigger) industry.  I get it.  I get why we do it and the need for it.

 

....but WAR is an (imperfect) tool, not the answer.  too many people treat it and the bastardized use of it, as the answer.

 

What we should be talking about is more along the lines of correct application.

 

fwiw, Fangraphs doesn't really want you (or me) to do that.  They want to put the disclaimers in the text and they put the number in a chart, because people are sheep and want to do what's easy.

 

Refining the stat is fine (good even)...but we should understand what we're refining it to do.

 

Why vote if the answer is WAR? (...wait...who's WAR?)

[CA-ORIOLE]

....in the sub title.

 

 

 

You think it's about Judge - Altuve, but it's not.  They are an example in the article, but my issue with WAR is the horrible mis-application of the stat and what it actually means.

 

I get metrics.  Someday maybe I'll share.  I have a 40M plan to address challenges that have never been addressed in a 30B (actually much bigger) industry.  I get it.  I get why we do it and the need for it.

 

....but WAR is an (imperfect) tool, not the answer.  too many people treat it and the bastardized use of it, as the answer.

 

What we should be talking about is more along the lines of correct application.

 

fwiw, Fangraphs doesn't really want you (or me) to do that.  They want to put the disclaimers in the text and they put the number in a chart, because people are sheep and want to do what's easy.

 

Refining the stat is fine (good even)...but we should understand what we're refining it to do.

 

Why vote if the answer is WAR? (...wait...who's WAR?)

1. Never said implied it's only about Judge/Altuve (mayne Macus beleves that) . Just the opposite. I believe it was largely a broader opinion/criticism about WAR and Jude/Altuve were merely examples that James had used to validiate that opinion. y. 

2. Never said or indicated that WAR is a prefect tool.

3. Never said refining the stat wasn't fine or that people searching to refine it should NOT know what they're doing. Most people would agree Bill James know what he's doing (even if you disagree with certain things). 

 

Do you have something to add that is relevant and/or not a strawman? 

[dude]

1. Never said implied it's only about Judge/Altuve (mayne Macus beleves that) . Just the opposite. I believe it was largely a broader opinion/criticism about WAR and Jude/Altuve were merely examples that James had used to validiate that opinion. y. 

2. Never said or indicated that WAR is a prefect tool.

3. Never said refining the stat wasn't fine or that people searching to refine it should NOT know what they're doing. Most people would agree Bill James know what he's doing (even if you disagree with certain things). 

 

Do you have something to add that is relevant and/or not a strawman? 

 

1.  Actually, you did.  I asked you what the article was about and while you were trying to be snarky and use your passive aggressive BS, you said it was about Judge/Altuve and James.  You guys went off trying to refine what we should be using in the MVP vote and what James thinks about it, blah, blah blah (no issues there, it's a discussion, but not the point of the article).  Again, the real question (and the one that Rany brings up) is how we should be using WAR.

 

I don't like quoting this much...I should probably include a couple more paragraphs too....

Over the last five years, WAR has caught on with the baseball-watching public in a way that no other sabermetric stat has. As Posnanski put it, “WAR has won.” It’s now on stadium scoreboards, and TV broadcasts, and in newspaper columns. We may not be that far from the day when the casual fan understands the meaning of a “five-WAR season” the way they now understand the meaning of hitting .300.

 

So it’s probably time for those of us who work in analytics to agree on the meaning.

 

WAR’s emergence as the gold standard of analytic stats was a happy accident. If it had been the result of meticulous planning, we wouldn’t have ended up with two competing entities each designing a metric with the same name that purported to add up every contribution a player made at the plate, in the field, on the bases, and on the mound. So instead of referring to “WAR” as a single product, we have bWAR (Baseball-Reference) and fWAR (and FanGraphs), the Coke and Pepsi of value-over-replacement statistics. (Baseball Prospectus has its own metric, wins above replacement player, or WARP, which hasn't achieved the traction of the other two.)

 

In 2013, Baseball-Reference and FanGraphs had a summit of sorts, in which they agreed on the definition of “replacement”: that is to say, where to set the baseline on the quality of a replacement-level player—the kind freely available in Triple-A or on waivers. But they did not agree on a formula to calculate WAR. That lack of consensus is a blessing in some ways; for an industry frequently accused of engaging in groupthink, there’s value in having multiple methodologies for trying to arrive at the same answer. What isn’t a blessing is the lack of consensus on what WAR is supposed to do.

 

2. Your first response in every discussion is (generally) he's a X WAR player.  You talk about changes to WAR value (for example, wrt Jones' defense) as if that's an actual thing.  It's your first comment about player performance or who a player is.  What were your first comments on Beckham?  You use it as your core analysis tool all of the time.  Don't hide behind the semantics of the word 'perfect'.  Nobody said you said the stat was perfect (that is actually a strawman).  If your focus isn't basically (only) WAR....what else do you use to evaluate a player? 

 

3.  Don't know why you commented on this to me, but ok.

-------------

 

No, I have nothing to add.  I really have no thoughts on anything.

 

....wait....how do you think we should use WAR?

 

What is your perspective on the best use?  ...the worst use?

[CA-ORIOLE]

1.  Actually, you did.  I asked you what the article was about and while you were trying to be snarky and use your passive aggressive BS, you said it was about Judge/Altuve and James.  You guys went off trying to refine what we should be using in the MVP vote and what James thinks about it, blah, blah blah (no issues there, it's a discussion, but not the point of the article).  Again, the real question (and the one that Rany brings up) is how we should be using WAR.

 

 

 

What is your perspective on the best use?  ...the worst use?

You never asked me any such question. I responded to your post to Stoner as Follows: 

 

"The article is primarily about Bill James proposal to re-allocate individual player WAR using team wins and run differential, for which he provides a basic framework, methodology and examples. While the entire article may reference tertiary aspects and background of WAR, James proposal is the primary emphasis. If you scroll down far enough you'll find the author essentially agrees with what some of us have said here, having some of the same criticisms. I don't think anyone, much less "everyone" has "just thrown out technical aspects of WAR" in response to and/or for the purpose of "ignoring the Authors good points", just that we've been generally focused to the primary emphasis of the article. 

 

FYI, I do agree with you that the author is pretty sharp,  but perhaps a bit long winded (but I'm guess you would find that a positive). 

 

Do you have an input as to James approach here? Or perhaps something else that you'd like to contribute that we all missed?" 

 

I told YOU what I thought the article was about. ZERO indication that it was only about Altuve/Judge or that this was about only the MVP voting. 

 

----------

 

When Mackus made a post indicating he thought it was about Altuve/Judge and the MV I respoded as follows: 

 

"Well, he didn't mention pyhag as I recall, but it was relationship/adjustment of individual player value (specifically WAR) based on actual team  wins and run differential (I get those are the 2 elements of pythag and James invented it etc). While Judge and Altuve (both potential MVPs) were the main example/focus (the day before the MVP voting), I didn't gather the approach should only be applied to potential MVP's. I assumed he was setting that up as more of a glaring example to support his theory."  

 

Dumb and ignorant is one thing, but dishonest is another. I'm done with you for now dude. I'd love to answer your what is WAR question. Kudos for asking something relevant, but quite frankly you're a waste of time. What would we discuss.: The defensive metrics being worthless and Adam Jones having as much range as any other CF in the league by your eyeball test (or something  else equally absurd)? WAR not capturing your hyperbolic valuuation of grit, leadership Synergy?  

 

Feel free to have the last word dude, otherwise I'll anxiously await your next verbal diarrhea dump.