Productive outs gets run out of town on a rail, and yet people still pretend that WPA means something? Why is this? Because WPA is "geeky"? It might be the single stupidest stat of all time. It's not even a stat, it's a way to follow the ebb and flow of a bseball game. People who use it to evaluate players are somewhere beteen Creationists and Flat Earthers in the continuum of gullible.

It's a better version of what RBIs are supposed to measure: What a batter actually contributed in that game, that at-bat. If Pujols hit like crap in high leverage one season and Holliday hit great, then Holliday did probably contribute more to his team winning that year, even though Pujols is still the better player, still a better bet going forward, and still probably more deserving to win MVP.

WPA is measuring a lot of luck and team context. But that's how the real world works.

Yes, I think that WPA is appropriate in arguments like this. Someone observes that AROD always choked in the clutch last year. Someone else documents that he did fine in the clutch last year, which certainly WPA can demonstrate.

Then the first person observes that AROD sucks, and the argument is over :)

Pujols hit like crap in high leverage one season and Holliday hit great, then Holliday did probably contribute more to his team winning that year, even though Pujols is still the better player, still a better bet going forward, and still probably more deserving to win MVP.


The way WPA measure high leverage is what makes it useless. If you hit a grand slam in the first inning and your team wins 4-3, then that was a very high leverage at bat in retrospect.

ALL RUNS COUNT THE SAME!

WPA=teh dumb

Let me put it another way, when a a stat says that the guy who hit a sac fly in the bottom of the ninth to win a 5-4 game contributed more to that victory than the guy who hit a three run homer in the first inning, that stat is useless in telling us anything.

I agree with that. And it can be fixed.

Productive outs gets run out of town on a rail, and yet people still pretend that WPA means something? Why is this? Because WPA is "geeky"? It might be the single stupidest stat of all time. It's not even a stat, it's a way to follow the ebb and flow of a bseball game. People who use it to evaluate players are somewhere beteen Creationists and Flat Earthers in the continuum of gullible.

First, productive outs mostly receive the reception they do, because the MSM completely overstates how valueable they are. I don't think anybody with a brain would rather have a productive out than a non-productive out. Just don't pretend that this is somehow on a par with the guy that got on base, or something like that. Also, as a measurement of anything it's pretty meaningless, because players that are productive without causeing outs don't rack them up as fast, different players have a different number of opporunities, and some players are asked to go for productive outs more often than others.

Secondly, WPA does "mean something". But you have to be aware of what it is that it's trying to measure. It isn't attempting to measure "how good is a player", and anybody that tries to use it that way is mistaken. But, when somebody asks the question "was player X clutch?", you can't awnser "he hit 330/400/600 with 42 HR". That's the wrong awnser, WPA does a pretty good job at awnsering that question (imho moreso than anything else I've seen).

Let me put it another way, when a a stat says that the guy who hit a sac fly in the bottom of the ninth to win a 5-4 game contributed more to that victory than the guy who hit a three run homer in the first inning, that stat is useless in telling us anything.

Would you accept a compromise in the form of re-naming WPA as "Fan Happiness Added?"

That's the wrong awnser, WPA does a pretty good job at awnsering that question (imho moreso than anything else I've seen).

No it doesn't.

WPA ignores all future events. It only has value in the instant the event occurs, since you can't tell the future. After the game is over, when you know the outcome, WPA is useless.

EDIT: WPA/LI might have some actual value though. Unfortunately the only person I have ever seen use it is Tango.

Bahaha. I like it, Eric J.

WPA should be called the Drama Index, since it effectively measures late, wild swings in victory probability.

But, when somebody asks the question "was player X clutch?", you can't awnser "he hit 330/400/600 with 42 HR". That's the wrong awnser, WPA does a pretty good job at awnsering that question

But it still doesn't answer why you would rate the player who hits the sac-fly in the 9th (in a 4-4 game), as more "clutch" than the player who hits the 3-run HR in the 1st (in a 0-0 game than ends up 5-4.

The 3-run HR seems WAYYYYY more "clutch" to me.

If WPA is the answer the question is wrong.

If WPA is the answer the question is wrong.

Bravo.

"What was the abbreviation for FDR's Works Progress Administration?"

Wrong question, ############.

WPA ignores all future events. It only has value in the instant the event occurs, since you can't tell the future.

If we're measuring how A-Rod hits in the clutch, I consider this an asset. A-Rod has no idea how the rest of the game is going to play out, so LI serves as a measure of the amount of "pressure," or whatever, that he's under in a given situation.

it still doesn't answer why you would rate the player who hits the sac-fly in the 9th (in a 4-4 game), as more "clutch" than the player who hits the 3-run HR in the 1st (in a 0-0 game than ends up 5-4

But that's precisely why WPA is a good tool in these stultifying AROD arguments. Great players (Schmidt, Bonds, AROD, whoever) are often criticized for hitting 3-run homers when the pressure is low, but never coming through with the little things that make all the difference. If you can show that they come through when the little things made all the difference, you can at least stop having that pointless argument, though since people are still compelled to write articles about it, perhaps not :)

are often criticized for hitting 3-run homers when the pressure is low, but never coming through with the little things that make all the difference.

Yeah, that's great. The problem is that a 3-run homer in the 1st with "no pressure" is worth the same as a 3-run homer in the 9th when both games end 3-2. WPA would say differently and that is why it is useless.

If you start the thread by calling the stat and anyone who uses it stupid, you don't deserve a discussion. I'm all for a discussion on the usefulness of WPA and any alternatives, but not when the waters have already been poisoned.

But that's precisely why WPA is a good tool in these stultifying AROD arguments. Great players (Schmidt, Bonds, AROD, whoever) are often criticized for hitting 3-run homers when the pressure is low, but never coming through with the little things that make all the difference. If you can show that they come through when the little things made all the difference, you can at least stop having that pointless argument, though since people are still compelled to write articles about it, perhaps not :)
Page 1 of 1 pages

Why counter dumb arguments with a junkstat? If a player goes through the season and has an MVP caliber year and for some freakish reason a large proportion of his production comes in the first four innnings or whatever, that player helps his team win way more games than the light hitting infielder who manages a mostly random collection of scrappy late inning RBIs. Giving your team a 3-0 lead in the 2nd inning is hugely clutch.

If you start the thread by calling the stat and anyone who uses it stupid, you don't deserve a discussion. I'm all for a discussion on the usefulness of WPA and any alternatives, but not when the waters have already been poisoned.

waaah waaaah

Did you make similar comments about the hundreds of simmilar comments made (justifiably!) about Productive Outs, or even RBI? WPA is so maddeningly stupid it really doesn't deserve a discussion of its merits. I think I've said it before, but it seems like it found favor with SOSHers and Ortiz fanboys in 2005 during MVP discussions and has managed to hang around ever since despite all its uselessness being exposed over and over.

If we're measuring how A-Rod hits in the clutch, I consider this an asset. A-Rod has no idea how the rest of the game is going to play out, so LI serves as a measure of the amount of "pressure," or whatever, that he's under in a given situation.

It's not a direct measurement, rather the measurement of a proxy for a direct measurement. I have my doubts as to how useful it would be at answering that particular question.

It just doesn't capture enough information and will never be able to. To really answer the question you need real observation data about the actual game event. Did he visually appear to be "pressing", was he in a funk anyway when put in the situation, did he get out on a nasty slider, did he hit a liner direct to an outfielder? Etc., etc., etc.

Did you make similar comments about the hundreds of simmilar comments made (justifiably!) about Productive Outs, or even RBI? WPA is so maddeningly stupid it really doesn't deserve a discussion of its merits. I think I've said it before, but it seems like it found favor with SOSHers and Ortiz fanboys in 2005 during MVP discussions and has managed to hang around ever since despite all its uselessness being exposed over and over.

I often wonder how the discussion of WPA would be different if Ortiz had been a Yankee and Arod a Red Sox, from both sides of the rivalry.

WPA does exactly what it says it does- measures each play's contribution to the team's chances of winning. Like it or not, that's something a lot of people are interested in. Heck, I'm interested in it too, even though I wouldn't use it to determine how good a given player is.

Or, perhaps I'm maddeningly stupid.

Why counter dumb arguments with a junkstat?

Well, I suppose one doesn't have to counter dumb arguments at all; one could just let them pass. But if you can counter an argument by showing that even in its own weakly conceived terms (e.g. "AROD always sucks when the Yankees need a base hit to win a game in the ninth") it's inaccurate, then why not?

It's not a direct measurement, rather the measurement of a proxy for a direct measurement.

Yeah, I realize that - it's what I meant by "serves as a measure." I had originally written "measures," and changed it to a weaker form on purpose. Mea culpa for ambiguosity.

No, it's not perfect. Still, if A-Rod leads off the top of the 9th with a game-tying HR, and the Yankees go on to score 6 more runs in the inning and win easily, I think most people would agree that the initial HR came in a clutch AB. Leverage and WPA give you that.

Also, if you select plate appearances in which A-Rod appears to be pressing, you may run into selective sampling issues - he might be more likely to press against a pitcher who's a bad matchup for him, for instance. He'd look worse in the "clutch" by that definition because it selects more difficult PA, rather than more important ones.

If you start the thread by calling the stat and anyone who uses it stupid, you don't deserve a discussion. I'm all for a discussion on the usefulness of WPA and any alternatives, but not when the waters have already been poisoned.

Seconded.

Except the question is odd and has very little to do with "clutch" (or rather it's apparent absence) and a whole lot to do with attitude and personal bias. You don't alter attitudes my providing the answer to the question you do it by demonstrating it was a bad question.

Also, if you select plate appearances in which A-Rod appears to be pressing, you may run into selective sampling issues - he might be more likely to press against a pitcher who's a bad matchup for him, for instance. He'd look worse in the "clutch" by that definition because it selects more difficult PA, rather than more important ones.

Which is something else you would have to consider when evaluating "pressing". It was not meant to be an exhaustic list of things. Merely a demonstrator that when it comes to an individual player and attempting to measure many single discrete events ignoring all these things is dangerous and can lead to a non-answer.

WPA does exactly what it says it does- measures each play's contribution to the team's chances of winning. Like it or not, that's something a lot of people are interested in. Heck, I'm interested in it too, even though I wouldn't use it to determine how good a given player is.

Or, perhaps I'm maddeningly stupid.

No, no, a thousand times no. It measures each play's contribution to the team's chances of winning at the time the play occured when you do not know what will happen in the rest of the game. When a game is over and you know the outcome, there is absolutely no value to WPA. It ignores information that you have after the game is over.

No, no, a thousand times no. It measures each play's contribution to the team's chances of winning at the time the play occured when you do not know what will happen in the rest of the game. When a game is over and you know the outcome, there is absolutely no value to WPA. It ignores information that you have after the game is over.

See; but to me, that seems like a cool idea. I like that. I like WPA when used on squiggly Fangraphs graphs that show how big the swing was when a team turned a one-out double play with a guy on third in the top of the eighth. I dig that.

I just wouldn't use it to decide who the best player was.

One of the DBacks blogs I read used to use those charts after the game for doing a play-by-play of each event. Adding a narrative. In that context it was good. Using it to assign value to a player, not so much.

So when a reliever comes in the game in the 9th inning, it makes no difference whether his team has a 1 run or 10 run lead. Only an idiot would think that varying game situations have different values to a team's chances of winning.

A good way to measure clutch performance, I think would be to determine, using Leverage Index, at what point an at-bat is "clutch", and at what point an at-bat is not at all clutch, no matter the outcome. Then compare a player's batting lines in the clutch, unclutch, and moderately clutch situations.

A player could be a terrible hitter in the clutch, but happen to get all his hits in situations with the highest possible WPA, and thus look like a great clutch hitter, even though he's batting 50 points lower without an increase in walks or isolated slugging in those situations.

See; but to me, that seems like a cool idea. I like that. I like WPA when used on squiggly Fangraphs graphs that show how big the swing was when a team turned a one-out double play with a guy on third in the top of the eighth. I dig that.

I just wouldn't use it to decide who the best player was.

I agree, I like the graphs that show the swings in win expectancy throughout a game. It should never be used as a measure of a player's value or contributions though.

See; but to me, that seems like a cool idea. I like that. I like WPA when used on squiggly Fangraphs graphs that show how big the swing was when a team turned a one-out double play with a guy on third in the top of the eighth. I dig that.

I just wouldn't use it to decide who the best player was.

Seriously. Fan Happiness Added.

A good way to measure clutch performance, I think would be to determine, using Leverage Index, at what point an at-bat is "clutch", and at what point an at-bat is not at all clutch, no matter the outcome. Then compare a player's batting lines in the clutch, unclutch, and moderately clutch situations.

It would be better than WPA but you'd also need a large enough body of "history" to ensure you not just capturing noise. If there was the data available a cross era comparison would be useful to see if you could get an idea of what changing bullpen roles had on this.

WPA does exactly what it says it does- measures each play's contribution to the team's chances of winning.

I was about to say NO IT DOES NOT... but #28 beat me to it.

See; but to me, that seems like a cool idea...

I just wouldn't use it to decide who the best player was.

I understand that, I even leaned towards that POV, and then I started seeing the mis-use of of WPA in MVP debates and such, the relentless mis use of a fun stat made it a stat with negative value imho.

RBI is a terrific stat, it is also over valued and mis-used by the MSM (and many fans) to such an extent that it's value is negated and then some. The problems with RBI
(non exhaustive list to follow)
1: Its adherents never seem to take opportunity into account- it means nothing to them than one guy had 300 baserunners available and another 200.
2: Its adherents never seem to take into account that high RBI/low OBP batters (Joe Carter anyone, Frenchy in his 100 ribbie year) are actually reducing everyone else's RBI opportunities- by using up outs and not getting on base- having Joe Carter drive in a large portion of the team's runs is actually a bad thing.
3: Its adherents really seem to believe that a guy with 100 ribbies produced more runs than another guy with 90 ribbies - even if the guy with 90 ribbies had 75 less PAs and made 100 fewer outs.
4: Its adherents are mortally convinced that Steve Garvey was a better hitter than Keith Hernandez.

Someone used a win expectancy-lite to determine that the 2003 postseason was the most "exciting" of all time.

Can't say I disagree. Two fantastically thrilling championship serieseses and, for a six gamer, a pretty solid World Series.

WPA does exactly what it says it does- measures each play's contribution to the team's chances of winning. Like it or not, that's something a lot of people are interested in. Heck, I'm interested in it too, even though I wouldn't use it to determine how good a given player is.

Some people are interested in this kind of thing. Some aren't. I don't understand why the people who are not interested insist on disparaging WPA. You're not arguing against anybody.

"The sky is clear"

"Who cares how clear the sky is?"

"Today is a nice day. It's 15% cloudy"

"Percent cloudy means nothing"

"It does if you want to measure how clear the sky is"

"It doesn't matter how clear the sky is. Therefore percent cloudy is meaningless"

"But I want to know how clear the sky is. So percent cloudy means something to me"

"It means nothing. It doesn't affect A, B, or C."

"But I want to know how clear the sky is"

"That's a stupid thing to want to know. Therefore the stat is meaningless"

First, productive outs mostly receive the reception they do, because the MSM completely overstates how valueable they are. I don't think anybody with a brain would rather have a productive out than a non-productive out. Just don't pretend that this is somehow on a par with the guy that got on base, or something like that. Also, as a measurement of anything it's pretty meaningless, because players that are productive without causeing outs don't rack them up as fast, different players have a different number of opporunities, and some players are asked to go for productive outs more often than others.

Secondly, WPA does "mean something". But you have to be aware of what it is that it's trying to measure. It isn't attempting to measure "how good is a player", and anybody that tries to use it that way is mistaken. But, when somebody asks the question "was player X clutch?", you can't awnser "he hit 330/400/600 with 42 HR". That's the wrong awnser, WPA does a pretty good job at awnsering that question (imho moreso than anything else I've seen).

You are right WPA has a small amount of value, but your first point is almost exactly in line with the way WPA adherents treat this gimmick stat. It's a small low important stat that is close to relavent as ops+ with runners in scoring position. It has value when treated as it should be, but that isn't the way it's treated. At no point should WPA ever be the first ten stat listed in an MVP argument yet it still gets trotted out there all the time. It's a complementary stat, that is it.


Heck I loved the concept of productive outs, but the execution was silly and not really a legitimate attempt to measure what they were claiming they were trying to measure. A tight focused stat has value when it's weakness's are properly acknowledged.

or I could have said, see post 36.

I think you're asking too much from WPA. It measures one thing, it measures it in an unambiguous way, and it happens to be something that people find interesting. That's enough for one stat. You're trying to say that it should behave like Win Shares or VORP or some other hero stat, and criticizing it when it doesn't. But Win Shares and VORP don't measure the only interesting thing in the world -- in fact, total contribution to wins is kind of boring.

I think you're asking too much from WPA. It measures one thing, it measures it in an unambiguous way, and it happens to be something that people find interesting. That's enough for one stat. You're trying to say that it should behave like Win Shares or VORP or some other hero stat, and criticizing it when it doesn't. But Win Shares and VORP don't measure the only interesting thing in the world -- in fact, total contribution to wins is kind of boring.

I think you have it backwards, people that tout WPA tout it as if it is a holy grail type of stat, they use it in situations that it isn't a good tool and act like it's the clear most important stat out there. As 36 says, RBI's has value when used properly, but because it isn't being used properly it is making a mockery of the stat, similar with WPA. If WPA is used in a similar vein as RBI's is intended and all the weakness's are acknowledged then it has value.

Personally I don't have a big problem with this article, but as a general rule, when WPA is used it always makes me check for the persons writing agenda. And it is almost always in some predetermined bias that is being used to prove their viewpoint on an issue.

No, no, a thousand times no. It measures each play's contribution to the team's chances of winning at the time the play occured when you do not know what will happen in the rest of the game.

I think we all understand that.

When a game is over and you know the outcome, there is absolutely no value to WPA. It ignores information that you have after the game is over.

But this is an entirely different statement. The fact remains that lots of folks find it interesting to know that the team had a 30% chance to win the game until A-Rod hit that home run, after which they had a 60% chance to win. I'm not saying it measures ability or even value, just that its interesting to me and others to see how much impact a given player had on changing the likelihood of winning. Just because you aren't interested doesn't mean other people can't or shouldn't be.

Just because you aren't interested doesn't mean other people can't or shouldn't be.

Or you could read my posts in the thread and notice that I do find the swings in win expentancy interesting. WPA is still useless though as it attempts to translates those changes in win expectancy into some measure of a player's value. Also, read post 41.

Seems we go through this every six months, and always on a Friday too.

As an added benefit to those who don't want to RTFthread, I basically agree with the following statements already made:

It's not even a stat, it's a way to follow the ebb and flow of a bseball game.
...
WPA is measuring a lot of luck and team context. But that's how the real world works.
...
Would you accept a compromise in the form of re-naming WPA as "Fan Happiness Added?"
...
It only has value in the instant the event occurs, since you can't tell the future.
...
WPA/LI might have some actual value though.
...
WPA does exactly what it says it does- measures each play's contribution to the team's chances of winning. Like it or not, that's something a lot of people are interested in.
...
It measures each play's contribution to the team's chances of winning at the time the play occured when you do not know what will happen in the rest of the game.
...
I like WPA when used on squiggly Fangraphs graphs that show how big the swing was when a team turned a one-out double play with a guy on third in the top of the eighth. I dig that. I just wouldn't use it to decide who the best player was.
...
One of the DBacks blogs I read used to use those charts after the game for doing a play-by-play of each event. Adding a narrative. In that context it was good. Using it to assign value to a player, not so much.
...
Seriously. Fan Happiness Added.
...
Some people are interested in this kind of thing. Some aren't. I don't understand why the people who are not interested insist on disparaging WPA. You're not arguing against anybody.
...
I think you're asking too much from WPA. It measures one thing, it measures it in an unambiguous way, and it happens to be something that people find interesting. That's enough for one stat.

I can sort of agree with this...

Let me put it another way, when a a stat says that the guy who hit a sac fly in the bottom of the ninth to win a 5-4 game contributed more to that victory than the guy who hit a three run homer in the first inning, that stat is useless in telling us anything.

...but that means then you don't give Mariano Rivera's 80 innings any more or less credit if they happened in the 9th inning of close games, than in the 7th innings of blowouts (that his team wins). So, you need to resolve how you can justify paying Mo and his ilk 10MM.

But it still doesn't answer why you would rate the player who hits the sac-fly in the 9th (in a 4-4 game), as more "clutch" than the player who hits the 3-run HR in the 1st (in a 0-0 game than ends up 5-4. The 3-run HR seems WAYYYYY more "clutch" to me.

...ONLY if you knew that the game would be 4-4 in the 9th inning. If you didn't know that, if there was a chance that the game would be 9-4 or 4-9, then your view of that 3-run HR would change. Your view of that sac fly would not change.

people that tout WPA tout it as if it is a holy grail type of stat

Those people are in the minority. They are not the ones to listen to any more than you would listen to people who scream insanity and say "God" at the end to mean that God-worshippers are morons.

I disagree completely with thise statement, among many others:

and has managed to hang around ever since despite all its uselessness being exposed over and over.

***

Larry/32: Sean has exactly that on his site. I gave him my LI for exactly that purpose.

WPA is still useless though as it attempts to translates those changes in win expectancy into some measure of a player's value.

But how does it do that? Just because it totals and we can rank it? No, the stat doesn't do any such thing. Some people may use it that way, but the stat does not. You're arguing that it doesn't do what you want it to do, not that it doesn't do what the people on this thread are claiming.

It's been stated that the 3-run HR in an eventual 4-3 victory is worth the same whether it comes during the 1st or 9th inning. That is true. And if we're talking value, you're right that WPA will overvalue the 9th inning one. But what kind of pressure to succeed do you think A-Rod is facing in those two situations. A-Rod comes up in the bottom of the 1st in a game where his team is already leading 1-0. Where does that fall on a pressure scale? 2/10? How will A-Rod approach this AB? Now take the 9th inning one, two outs, down 3-1, two men on base. What's that? 9/10? Will A-Rod approach that AB differently? If you're trying to measure whether A-Rod is clutch, a stat that treats those situations differently is essential. Because if A-Rod is a choker, he'll hit more HR in that 1st inning situation than in that 9th.

Let me also reprint something I've written in the past:

As for the HR in the 1st or 9th: the assumption of win expectancy, WE, (as in life) is that everything in the future is unknown, and therefore, your expectancy is that average things will happen. So, hitting a HR in the 3rd increases your win expectancy by, I dunno, +.130 (say from .500 to .630), because you have a certain expectancy of what will happen in the following 6 innings. So, the WE represents how much you are willing to bet on the game’s outcome at that point in time, and the change in WE (i.e., WPA) simply pays off on that bet.

In the 9th inning, you are “all in”. You may have say won most of the poker hands before that, but if someone goes all in, and he wins, you lose.

That’s all WPA is… an accounting of money. In real-life terms, as studes put it, it simply tracks your emotions, and quantifies it.

So, I can see why WPA gets a bad rap… either because it is misunderstood or misapplied. OBP doesn’t get a bad rap, even though it weights a walk and HR as exactly 1. Its limitations are self-evident, and therefore, no explanation needed. With WPA, we have to work harder at explaining what it is and what it isn’t.

Belittling WPA, as a stat, makes no sense, because it represents what it purports to represent. Nothing more.

...but that means then you don't give Mariano Rivera's 80 innings any more or less credit if they happened in the 9th inning of close games, than in the 7th innings of blowouts (that his team wins). So, you need to resolve how you can justify paying Mo and his ilk 10MM.

You're correct. I give closers very little credit for the leverage. If I were a GM, I don't think I'd ever pay a reliever anywhere close to $10 mil.

To me a great closer ranks somewhere around a decent 4th starter in value. I think an Andy Pettitte season ending injury would hurt the Yankees far more than losing Rivera. (Except, perhaps, for the idiotic decision of making Joba the closer that might follow). But if you let Bruney be the closer, I'd rather lose Mo than Pettitte.

You're correct. I give closers very little credit for the leverage. If I were a GM, I don't think I'd ever pay a reliever anywhere close to $10 mil.

That's fine then. You are consistent in your position, that having a 2.50 ERA reliever for 81 innings at LI = 2 is really no different than a 4.50 ERA starter for 162 innings.

If you set the replacement level at 4.50 for relievers and 5.50 for starters, then your reliever is 18 runs above replacement, just as the starter is.

Such a fantastic reliever and below-average starter would make 8MM on the open market (pre-economic doom), according to your position.

You would be in the minority in your position. Not to say you are wrong, but at least you are consistent.

Monty/49: you are technically right.

There are nearly 200,000 events every year. LI will tell you in 30 seconds the quick way what would take you 200,000 minutes the right way, with 90% overlap in conclusion. Is that really something worth arguing?

You may as well argue that "hits" is not a good enough indicator, because someone may have the bad luck to hit a disproportionate number of his great swings when Endy Chavez is fielding. Yes, we are better off counting each at bat one at a time to see how many hits he should have gotten had he faced a normal distribution of fielders. That's alot of work. That's why we live with UZR (and PZR, HZR).

Clutch, at least as it is defined by the media and a good portion of the warblers, is a case of the ever moving goal posts. It changes from player to player and situation to situation. Which is in part why I think expending effort attempting to answer that question is forever doomed to confusion, changeable defintions and misinterpretation of the results of what effort is expanded. Of which the misuse of WPA is an example.

Change the question.

As for closers. I have no problem understanding that their might be some psychological benefite to a Mariano Rivera but in terms of measurable value I generally agree with Snappers take. With the caveat that I think a bullpen is important while the leverage of a situation a reliever finds himself in is of less importance than the level of ability that reliever has.

That's fine then. You are consistent in your position, that having a 2.50 ERA reliever for 81 innings at LI = 2 is really no different than a 4.50 ERA starter for 162 innings.

If you set the replacement level at 4.50 for relievers and 5.50 for starters, then your reliever is 18 runs above replacement, just as the starter is.

Such a fantastic reliever and below-average starter would make 8MM on the open market (pre-economic doom), according to your position.

You would be in the minority in your position. Not to say you are wrong, but at least you are consistent.


But actually, they're pretty close. Rivera got his $15M b/c of the "special Yankee legend bonus". No one else was going much above $10M.

In the last couple of years your Jeff Suppan's and Francisco Corderos were both going for about $10M per, with the SP getting longer deals.

We also saw a big decrease in closer demand this year. I think we're withing spitting distance of the ace RP and good #4 SP getting similar money. Pettitte and Fuentes will likely make about the same this year.

The other thing about WPA is that it adds up to actual wins. Ordinary stats add up to Pythagorean wins or component-based wins ("first-order" or "second-order" by BP's terminology). The only exceptions are pitcher's W-L, Win Shares, and WPA. (Am I missing anything? I guess GWRBI would also count.)

I agree with the objections in this thread. But, what if a team beats its pyth record by ten games and makes the playoffs, and there's an MVP campaign for its top hitter? The writers are going to say he carried the team on his shoulders. His stats may not look as good as Pujols or Sizemore's, but that's why they call it MVP and not the Best Player Award. Blah blah blah, we've heard it all a thousand times.

How do we respond? We could say, that's a misguided view. It should be the Best Player Award. Okay, but that's purely a matter of opinion. Not much of a response.

We could say, it's futile to divide up credit for those extra ten wins among the players. WPA does it one way, which has clear flaws. Win Shares splits the credit in proportion to basic offensive contribution, which might satisfy some of us and not others. We'll never all agree on a single method. Still, saying to hell with it all and going back to VORP is pretending that the extra ten wins didn't exist.

I'm in the process of creating a new stat similar to WPA, in which the credit for each event depends only on:

* the run environment
* which team was batting
* the base/out situation before the event
* the base/out situation after the event, and the number of runs scored during it
* the final score of the game and the number of innings (adjusting for partial innings)

So, if the visiting team wins 5-3 in 9 full innings, a one-out double with a man on 3rd hit by the home team has the same value no matter which inning it occurred in and what the score was at the time.

As with WPA, the winning team ends up with +0.5 wins and the losing team ends up with -0.5 wins.

I'm interested in hearing what you all have to say about this. The concept certainly isn't perfect, let alone the implementation. My own view is that this stat isn't any more or less inherently valid than WPA. It's simply a matter of taste. For MVP discussions, the idea that there's one true stat which everyone should follow is just wrong. So there are multiple stats (and non-statistical things) to consider, which everyone can weight as they like; and I hope that my work-in-progress, when it's finished, can join that group.

I actually like to use WPA as a sanity check. (As is the case in the article that's linked to)

It's also interesting to look at when a team misses its pythag by a big margin.

All in all I think it adds to the narrative while having no true significance about how good any given player is. Never seen anything to make me think it has any predictive value. Even for relief pitchers -- where I think it brings something to the table.

But actually, they're pretty close.

The numbers I gave, those specific numbers, is 18 runs above replacement.

If you used leverage, you'd give Mo an extra 50% and he'd come in at 12MM$, which is what any team would pay a reliever with a 2.50 ERA.

The 4th starter gets 8MM, as is standard.

The specific examples you have have to be interpreted based on their exact wins above replacement. We were just using illustrations. I've done all the work already, and relievers are paid as per my model (LI/2+0.5), as are all the starters paid as per my model.

We've agreed that you are consistent. Let's agree that you are in the minority, and we'll call it a day.

I think we're withing spitting distance of the ace RP and good #4 SP getting similar money

Remember, I said ERA of 2.50. An ace reliever is nowhere that low (in terms of true talent). I took the best possible reliever you could hope for, and compared him to the #4 starter. So, we're not there, and we're not going to be there.

when a team misses it's pythag by a big margin one of the first things I look at is the bullpen and flakey starting pitching rather than the hitters.

And I like the idea of determing the value more by context than inning sounds better than WPA at what spme people use WPA for even if it ends up as a model for something different.

I can sort of agree with this...

Let me put it another way, when a a stat says that the guy who hit a sac fly in the bottom of the ninth to win a 5-4 game contributed more to that victory than the guy who hit a three run homer in the first inning, that stat is useless in telling us anything.

...but that means then you don't give Mariano Rivera's 80 innings any more or less credit if they happened in the 9th inning of close games, than in the 7th innings of blowouts (that his team wins). So, you need to resolve how you can justify paying Mo and his ilk 10MM.

I'd say that's a false choice. You can believe that good performances in close games are worth more than good performances in blowouts, but still believe it doesn't matter in what inning the performance came (or the score at that time). If the Cards go on to win by 1 run today, Ludwick's HR is no less important in the 3rd inning than had it happened in the 8th (but WPA says differently). But for Mariano, pitching in a close game invariably means the game ended close, because of when he pitches. So I can believe that Mariano's performance has more than average value, but still reject WPA as a measure of value in many other circumstances.

Take Carpenter's first game last week: 7 IP, 1 unearned run, and the game was tied or within 1 run for every one of his pitches. Yet he gets an LI of just 1.02, and the 2 relievers who pitched the 8th and 9th end up with a higher LI and more WPA (between them) than Carpenter. That tells me something about who "really" contributed to this win? I think not.....

So, if the visiting team wins 5-3 in 9 full innings, a one-out double with a man on 3rd hit by the home team has the same value no matter which inning it occurred in and what the score was at the time.

Talk to studes. He wrote a mini-article several months ago where he handled that very thing.

That tells me something about who "really" contributed to this win? I think not.....

I'm not saying that it does. If someone is, then they are wrong.

Does someone with a .500 OBP contribute more than someone with a .250 OBP, if the first guy gets 2 IBB, and the second guy hits a grandslam?

WPA needs to be interpreted.

First of all, here's the game in question:
http://www.fangraphs.com/wins.aspx?date=2009-04-09&team=Cardinals&dh=0&season=2009

WPA is +.25 for Carpenter (7 innings), and +.285 for the other 2 relievers (2 innings).

There are 2 necessary adjustments:
1. The baseline for "value" should be replacement-level, not average.

2. I give a leverage factor that is halfway between whatever the LI is and 1.0

So, in step 1, we need to give all starters an extra +.013 wins per inning pitched to get it to the replacement level, and add +.003 wins per relief inning. That puts Carpenter at +.34 and the two relievers at +.29.

Secondly, their LI was 1.87, so we need to take out 43% of their original WPA of +.285, or .12. That puts them down to +.17.

The story here, in terms of "value" is +.34 for Carpenter's 7 innings, and +.17 for the reliever's 2 innings.

Like I said though, I wouldn't go with WPA on a value-perspective. That's not what it's there for. But you can try to use it with a good amount of adjustment.

Remember, I said ERA of 2.50. An ace reliever is nowhere that low (in terms of true talent). I took the best possible reliever you could hope for, and compared him to the #4 starter. So, we're not there, and we're not going to be there.

Ah, OK. I thought you were talking standard variety closer. Yes, a Rivera/KRod type is getting more. I wonder if that will persist.

Ludwick's HR is no less important in the 3rd inning than had it happened in the 8th (but WPA says differently).

But nobody said that WPA measures "important", either. You're saying that. It's a higher number, sure. And most people would say that had the HR happened in the 8th it would be more clutch. So people say 8th inning homeruns in a 0-0 game are more clutch. And WPA has that with a higher number. So what's the problem here? Are you arguing against what I just said, or some argument about value that nobody on the "other side" has made?

Fret, in this model you are working on, how would you credit batting events for the team that winds up losing?

Greg: I hadn't read your post, so definitely wasn't arguing against you. However, I don't think it's unfair or inaccurate to suggest that some WPA fans really do see it as an important measure of value. They will acknowledge that it can't measure talent, but say it does measure what "really happened" -- and thus value -- even if that was partially a function of luck. Here's Sky Andrecheck in his recent article on "Championship WPA." (He's talking about championship WPA here, but I've seen the exact same case made on behalf of WPA.)

Champ WPA of course, is also not predictive, and thus is not very useful in player evaluation, but it does measure the actual impact that a player did have in terms of championships. I think that's a fairly noteworthy thing to keep track of, if just from a historical perspective. Francisco Cabrera, in one swing, contributed greater share of a championship (37%) than thousands of better players did in their entire careers. It doesn't mean he was better, it just means he really did contribute more - even if only by luck.

I just think this view of WPA is factually wrong. It doesn't tell us what the "actual" impact of anything was. It tells only how we felt about what happened at that moment in time. Interestingly, this seems to be more obvious to people in the championship context. Everyone agrees, I think, that a win in April has exactly as much value as a win in September, exactly as much impact on whether a team wins or loses their division. Yet some still cling to the idea that a shutout inning in the 8th has more value than one in the 2nd.

I like the idea of creating a "retrospective leverage," that measures the importance of events once we know the outcome of the game, as Fret seems to be working on. Not sure if it's possible, but an interesting idea.....

What's more clutch? To hit a 2-out single with the count at 0-2 with a runner on third, or hitting a HR in the top of the 8th with no one on and no one out.

(Tie game, both)

Tango: I don't think we disagree very much (probably a little). I just think there's an important distinction between how close a game was at it's conclusion, and how close (and late) it was when an event occured. It's possible to think that one is important and not the other. In other words, you can think LI tells us something important about relievers, while also thinking WPA has little to tell us about other players.

The problem is that a 3-run homer in the 1st with "no pressure" is worth the same as a 3-run homer in the 9th when both games end 3-2. WPA would say differently and that is why it is useless.

They are worth the same runs, but not wins. That's why WPA is correct. The 3-run homer in the 1st stands a substantial chance of ultimately being entirely irrelevant to the game outcome. The homering team could go on to outscore the other team in the remaining innings anyway, or the victimized team could come back by a margin of 4+ runs. In either of those scenarios, the first inning HR becomes immaterial to the outcome. There is no such possibility for the ninth inning HR since it immediately ends the game and sends win chance to 100%.

What the above argument omits is the scenarios where the ninth inning HR has near-zero impact on winning percentage. If the HRing team is already ahead, or trailing by more than 3, that HR's impact towards winning is minimal.

A first inning 3-run HR has X chance (let's say 40%) of holding up as the winning margin throughout the game. A ninth inning 3-run HR also has exactly X average chance over all games to be the winning margin. That chance is just quantized into immediately visible near-100% or near-0% scenarios. We can't ignore the near-0% occurrences.

WPA makes a distinction between all of these scenarios and captures it numerically. Each HR is worth 0.4 WPA on average. The first-inning HR is worth 0.4 WPA every time since we don't yet know what will happen in the rest of the game. The last-inning HR is worth near-1 WPA 40% of the time, and near-0 WPA 60% of the time (ignoring tie score scenarios, but those average out to 0.5 WPA on each side. ) They are worth the same win probability when averaged over all instances of occurrence, but not the same win probability in each individual instance.

Put more concretely: Frankie Firstman hits 100 first inning HRs. Frankie receives 0.4 WPA for each HR or 40 WPA total. This comes from the fact that in ~40 of those games, the HR will stand up as the winning margin, while ~60 of those games will be lopsided enough anyway that the HR won't matter. Larry Lastman hits 100 last inning HRs. In ~40 of those games, the score will be close enough for that HR to be the winning margin and Larry receives 40 WPA. In ~60 of those games, Larry's team was already ahead or trailing by too much to matter and Larry receives 0 WPA. Both hitters got 40 WPA total. Now, if Larry's HR distribution came differently than expected -- say 75 of them came with his team closely trailing -- WPA would capture that and reward Larry with 75 WPA.

After the game is over, when you know the outcome, WPA is useless.

That's only true to the degree you don't think the end of the game is the sum of a bunch of prior events. I don't think that's the case, personally

WPA does exactly what it says it does- measures each play's contribution to the team's chances of winning.

GuyM has demonstrated rather incontrovertibly that this is false. If WPA actually did this it might be worth something.

That said, I agree with fret's comment that if you are asking who is most valuable you have to account for actual wins.

In other words, you can think LI tells us something important about relievers, while also thinking WPA has little to tell us about other players.

True enough. But you can use WPA - or as I prefer to do, LI - to provide an analytical framework for discussions of "clutch" performance, because WPA and LI mirror the fan's perception of relative game importance very well. When you start from the viewpoint that the 3-run HR in the first is just as important as the 3-run HR in the ninth, the discussion becomes a religious argument, not an analytical discussion. When I have presented LI to people who aren't into the heavy math, they get it - I show a list of typical LI 1.0 situations, LI 2.0 situations, LI 3.0 situations, and they understand it. So that when I then talk about performance in LI 2.0 and above, they know more or less instinctively that I've captured most of what they consider to be "clutch" - and then we can have a reasoned discussion about it.

As I have said before, I think there is tremendous value in using analytical tools to engage people in discussion about "their" views and beliefs about the way that the game is played, rather than using them to push people toward "our" views and beliefs about the way that the game "should" be played. Personally, I learn a lot more that way.

-- MWE

FWIW, here is the thought exercise I present to the people who do not understand the value of information or exhausting event spaces.


You are the GM of a team, and you get visited by your God of Luck. He grants your wishes, but he's a libertarian sort so he demands money to do it. You never know when he is going to come.

He shows up on the first inning on a getaway day at Kansas City during May. He asks you, "How much will you give me for a 3 run homer?" -- what do you answer?

He shows up again in the first of November, during the 9th inning of game 7 of the WS in a tie game, with a runner on third with one out. He asks you, "How much will you give me for a sacrifice fly?" what do you answer?

Talk to studes. He wrote a mini-article several months ago where he handled that very thing.

Thanks for letting me know. I wouldn't want to reinvent the wheel. I couldn't find the article on THT -- am I looking in the wrong place?

--

Fret, in this model you are working on, how would you credit batting events for the team that winds up losing?

The short version is that there's a right way to do it, and I do it that way. Read on for the long version.

To simplify things, ignore the issues of park factors, bias caused by the home team not batting in the bottom of the 9th, etc.

Then what we want is a formula that says, you lose 10-8, your offense gets +x wins and your defense gets -y wins, where x-y = -0.5. To make things symmetric, if you win 10-8 your offense should get +y wins and your defense should get -x wins, y-x = +0.5.

In other words, we need a function F(RS,RA) that gives the number of wins to credit the offense when you score RS and allow RA, such that
F(RS,RA)-F(RA,RS) = +0.5 if RS>RA and -0.5 if RS<RA.
In the example above, F(8,10) = x and F(10,8) = y.

The problem is that there are lots of possible functions F. For example, you could decide to let F(RS,RA) = +0.25 if RS>RA and -0.25 if RS<RA. That means, when your team wins you always give +0.25 wins to the offense and +0.25 wins to the defense. When you lose, both the offense and the defense get -0.25 wins. Obviously this seems wrong.

Suppose the average team scores (and allows) 750 runs per season. My team scores 900 runs and allows 950. It hits its pyth record on the nose, so it goes 76-86 (roughly, using 10 runs per win). Using the "wrong" function F, we'd give the offense -2.5 wins and the defense -2.5 wins. It makes more sense to give the offense +15 wins and the defense -20 wins.

This reasoning leads to a criterion for the function F. If a team's RS/RA distribution is typical, and we add up the total credits game-by-game for the offense and defense, the results should agree with the 10 runs per win rule (or more precisely, with the pyth formula equivalent).

It turns out (I believe) that this criterion determines F uniquely. So, that's what I use.

The gory details involve Weibull distributions (following work done by Steven J. Miller).

I think there's a simpler way involving Pythag, fret. Let's ignore for a second the revisions like Pythagenpat, and just stick with:

RS^2/(RS^2+RA^2)

For ease of computation.

A team that scores 900 runs would, with an average defense/pitching, win:

900^2/(900^2+750^2)=.590, or about 96 wins

A team that allows 950 runs while scoring the average, win:

750^2/(750^2+950^2)=.384, or 62 about wins

That gives +15 offensive wins and -19 defensive wins, same as your method.

This is the studes article:

Maybe you don't believe there's a difference between a first-inning home run and ninth-inning home run in the same game, but I believe there is definitely a difference between a home run in a one-run win and a home run in a ten-run win. In general, hits in close games have a bigger impact on winning than hits in blowouts.

To prove the point, I pulled all the WPA events from 2006 and calculated the average impact of each event on the win probability of the batter's team. Not surprisingly, I found that the same type of event (such as a single) in close games has a larger win impact than that type of event in blowouts. Based on the data, I have estimated a standard multiplier for calculating the impact of a hit or out in a game with a particular victory margin. It's still under development, but I don't think the final version will look much different than this:

Margin Impact
1 1.38
2 1.13
3 0.97
4 0.86
5 0.76
6 0.66
7 0.63
8 0.57
9 0.51
10 0.47

You might call this a "Game Leverage Index," as it shows that a single in a close game is worth about 40% more (in wins) than a single in a three-run game (1.38 compared to 0.97). And it's worth nearly three times as much as a single in a game with a 10-run margin (1.38 vs. 0.47). The same multiplier can be applied to any type of event.

I was actually surprised that the differences weren't greater, but this table makes a lot of sense to me. I think it could represent a workable compromise between those who want to value performance in games, but don't believe that the difference between "when" an event occurred should matter once the game is over.

I really like the idea behind it. It is similar to James' "Victory-important RBI" or whatever he called it, where an RBI in a blowout counted less than an RBI in a close game. What he did was if you gave up 3 runs, then you needed 4 runs, so you proportion all the RBI the winning team got down to 4.

Studes' concept at least satisfies the condition that a run in the 1st counts as much as one in the 9th, while at the same time making sure that ALL runs scored in a close games count more than all runs scored in a blowout.

I'm as sick as anyone of these WPA threads, but I've got to add a couple of things. First, I'm probably as big a fan of WPA as anyone else (including SoSH posters, etc. etc.) but I haven't claimed it is the "ultimate" stat. Tango hasn't either. People need to stop ranting at something that doesn't exist.

I am fascinated by WPA because it does things no other stat does. For instance, it serves as the basis for Leverage Index, perhaps the best sabermetric concept of the past 15 years. Also, the article that Tango linked to. Or the "heartburn reliever" article I wrote last week for THT downloads. And I'm absolutely flummoxed by people who don't see the value in that stuff.

People claim that it gained lots of traction from places like SoSH, but it actually took off here, when Tango had his blog here. He proposed it as an alternative to Win Shares.

We've used WPA at THT for several years. The past two or three years, I've used it to anoint the best "clutch" hitters of the year. I pull all plate appearances with an LI over 3.0 and basically figure runs created, OPS, whatever. WPA is a decent shorthand for the same thing.

I also invented the graphs that track WPA during games, and I happen to think it's my best contribution to the sabermetric field. I love watching those graphs progress during a game while also watching it on TV.

My bottom line is that WPA is an amazing baseball statistical tool for many things. John's article is just one example. Sit back and enjoy it.

Sit back and enjoy it.

Or if you provide neither constructive criticism nor debatable points, get out of the way.

Summary conclusions without evidence is not the way to stop the spin, regardless of what you hear on Fox "News".