I agree with Doc Nabbit about the SkyDome (Centre Rogers) being superior to Busch III.

I'm sorry I couldn't go this year, fellas. Alejandra could have played the piano for you guys - if they'd have let a 10-year-old in the bar. It's still school term here in the UK, and it's been the worst year on record, workwise, for me. Four transAtlantic air tickets don't come cheaply.

Theoretically, I'm moving to Canada in August, so maybe DC will be good for me, as I won't have to take all the family to see Grandma, too!

Oh, I forgot - I adapted the Game Score to help estimate the contributions of cricket players and it worked very well. I think Angus's views about the Game Score are spot on.

Mike,
a) sure, b) not IMO, c) even if they aren't, it's hard to imagine the sample size is going to get there even with differnet techniques.

I may come up next year. Mainly to play poker with the drunks.

When Repoz ordered it, I said "don't tell me that, you're on drugs!" And he said "No Craig, I'm not on drugs I'm okay, I was just thinking you know, why don't you get me a Pepsi." And I said, "NO you're on drugs!" But all he wanted was a Pepsi, and I gave it to him.

Doesn't matter. He'll probably get hit by a car anyway.

I'm not crazy. You're the one that's crazy.

Oh, and I saw your Mommy and your Mommy's dead.

Going from memory, Glavine has had years in which his RISP splits were outstanding, but not so much over his entire career. When they cracked down on his strike zone, I think it hurt him quite a bit.

it's hard to imagine the sample size is going to get there even with differnet techniques.

If you define clutch using leverage - which makes the most sense - the sample size isn't that small.

-- MWE

Going from memory, Glavine has had years in which his RISP splits were outstanding, but not so much over his entire career. When they cracked down on his strike zone, I think it hurt him quite a bit.

Yes, while talking to Cramer about the clutch pithcing I cited Cone, but looking back, not so much. Of course, the Cone I was thinking about was in 1998 or so, not the rest of his career.

If you define clutch using leverage - which makes the most sense - the sample size isn't that small.

Maybe not, but I'm not a big "early game" leverage fan.

Oh, I forgot - I adapted the Game Score to help estimate the contributions of cricket players and it worked very well. I think Angus's views about the Game Score are spot on.

fra - you can e-mail Jeff Angus (his e-mail is listed on the SABR directory on the members-only part of the website) and he'll give anyone who wants an 18 page paper that his presentation was based on. Or, if you'd rather, I can forward you the copy he sent me.

I may come up next year. Mainly to play poker with the drunks.

The Will Young theory.

Will and Aaron actually knew some other younger guys from Minnesota like Kerry and Kyle. I think I'll start recruiting some younger guys for Connecticut. There's only two or three that are younger than me who attend the local meetings.

Maybe not, but I'm not a big "early game" leverage fan.

What does that mean?

Oh, I forgot - I adapted the Game Score to help estimate the contributions of cricket players and it worked very well

linky?

I like the idea, Mike. Back of the envelope calculations:
what are 3 SDs from average, translated to OPS+? A 67 or a 133?

Using Scott Fletcher (the one they said was tops in "clutch"), he has a 120 tOPS+ with RISP, and a 117 with Men On. In "high leverage", he has a 114 tOPS+. he does have 1000 PAs with RISP.

Similarly, with Hidalgo, he has a 80, 88 and 65 in those same situations.

So, where are the bounds of competence? In HiLev, Hidalgo has 850 PAs. How many more runs would he account for with a 100? It is more than 10. Well, using PA*OBP*SLG yields about 50 runs difference for Hidalgo. They didn't say how much "un-clutch" Hidalgo was.

Doing the same for Fletcher, I get 17 extra runs created.

What does that mean?

I think decision-making changes it more than an average would indicate.

I think decision-making changes it more than an average would indicate.

This was meant to explicate your original statement? Remember who you're talking to and speak English, Busey.

Uh, in the early part of the game, from what I have seen, leverage is too generous. there are still too many human mistakes to make.

I like the idea, Mike. Back of the envelope calculations:
what are 3 SDs from average, translated to OPS+? A 67 or a 133?

I'm about as much of a math guy as Chris Jaffe, but I believe that the standard deviations change from year to year with these relative stats (OPS+ and ERA+) depending on the spread of talent.

I'm about as much of a math guy as Chris Jaffe, but I believe that the standard deviations change from year to year with these relative stats (OPS+ and ERA+) dpending on the spread of talent

Slightly, sure, but "oh so" slightly.

Oh, I forgot - I adapted the Game Score to help estimate the contributions of cricket players and it worked very well

linky?

You'll find some stuff at my now-defunct Sabermetric Cricket site: http://www.cathar.demon.co.uk

However, most of the actual Batsmen Score and Bowlers' Score is tucked away on various spreadsheets I have archived on CDs here and there. I also lost my most recent data when my iBook's hard drive gave up the ghost for the second time. I'd have to double-check whether I archived it in time.

Uh, in the early part of the game, from what I have seen, leverage is too generous. there are still too many human mistakes to make.

Thank you. Now that I know what your position is I can say I agree with it. I would probably have argued that leverage formulas (that I recall, mostly from 10 years ago on Usenet and probably out of date) tend to over-emphasis the value of late-inning events and under-emphasis the importance of early-inning events. I think that is what you're saying. If so, we have one of those rare occasions where Sam and Dial agree and thus the debate is probably over.

over-emphasis the value of late-inning events and under-emphasis the importance of early-inning events. I think that is what you're saying. If so, we have one of those rare occasions where Sam and Dial agree and thus the debate is probably over.

I agree. However, I think while that's true relative to one another, I htink it overvalues both.

I would probably have argued that leverage formulas ... tend to over-emphasis the value of late-inning events and under-emphasis the importance of early-inning events.

It sounded to me like he was saying the opposite, which means that he does agree with you in all likelihood.

Maybe not, but I'm not a big "early game" leverage fan.

Virtually all high leverage situations occur after the 5th.

-- MWE

Virtually all high leverage situations occur after the 5th.

Which seems very wrong to me.

I would probably have argued that leverage formulas ... tend to over-emphasis the value of late-inning events and under-emphasis the importance of early-inning events.

If you are using raw WPA, yes. But Tango's approach scales the values back to the average WPA swing per plate appearance, which gives you a more workable set of numbers.

I think that when you (or anyone) say leverage "over-emphasizes" and "under-emphasizes" certain PAs, you are arguing from your own perspective of how valuable those plate appearances "should" be, and if you are going to do that, it would help if you would describe a model of how you determine the importance of a plate appearance. Tango did that, and while that model has its flaws it has the virtues of being well-understood and roughly in agreement with most people's perception of the relative importance of game situations. I work with Tango's model because it is (IMO) the best available. If you have disagreements with the way that Tango weights things, it would help me a great deal if you'd be explicit about those disagreements and come up with a way to better measure game importance.

-- MWE

Which seems very wrong to me.

Why?

-- MWE

I suspect the answer to #127 is related to my potential reply to #126. That, in turn, will require me to do a little research into Tango (and thus your) weightings. I will probably disagree with that model assuming it weights events and outcomes based on how many innings are left in the game. I haven't thought it through in years - literally - but I'm not sure I believe an event/outcome in the first is more important than that same event/outcome in the ninth. But I can't put that conversation on paper until I see what you're working off of. Got a link to Tango's model? Preferrably one that involves as much of the grammar section of the SAT as the math section.

Doesn't Glavine have huge clutch type splits? I know there was something about him that was interesting...

I don't know enough about the over-all numbers to know if this is unusual or not, but one of the more notable things about Glavine's "clutch" is that in 425 plate appearances with the bases loaded, he's given up only two home runs. And one of those grand slams was hit this season in one of Eric "Blind Squirrel" Byrnes' few moments of competency.

Sam: Tango wrote up LI for THT a while back:

Part 1
Part 2
Part 3

I haven't thought it through in years - literally - but I'm not sure I believe an event/outcome in the first is more important than that same event/outcome in the ninth.

Not in retrospect. But in real-time, I think the crowd/manager and players all perceive things (and act) differently later in the game. If they didn't, we wouldn't have closers.

Will and Aaron actually knew some other younger guys from Minnesota like Kerry and Kyle. I think I'll start recruiting some younger guys for Connecticut. There's only two or three that are younger than me who attend the local meetings.

To be fair, they were already coming every year without being recruited (Kerry won the individual trivia contest in Cincinnati). I just invited them to tag along with us whenever they wanted and told them about the impending whiffle ball game.

Poker Friday night ended up being pretty fun. At 4am I started playing 5-10 heads up with some stranger in the hotel lobby and played until 6ish.

The entire week was fun. I wish I had more time with you all. Like usual it was a ball.

But in real-time, I think the crowd/manager and players all perceive things (and act) differently later in the game. If they didn't, we wouldn't have closers.

Exactly. You'd also see teams pinch-hit more often in the early innings, which they don't do very often.

Something for Chris J: Paul Richards pinch-hit for position players in the early innings a *lot* while he was managing Baltimore. The Orioles were in double-digits in that category - and had more than any other team in baseball - in every season from 1956-1961. No team has been in double digits since the Tigers in 1987.

-- MWE

Repoz Posted: June 30, 2008 at 02:14 PM (#2837618)

Did you really do this weekend sober?!

and the last 510 weekends.

- good for you there boy!!!! you almost at your 10 year pin!!!

AWESOME!!! dontchu listen to none of these guys. you rock!!!!!!!

- good for you there boy!!!! you almost at your 10 year pin!!!

AWESOME!!! dontchu listen to none of these guys. you rock!!!!!!!

Yeah, but think how much more awesome the lead-ins would be...

- good for you there boy!!!! you almost at your 10 year pin!!!

Thanks, BC, but no pins...cold turkey all the way.

Needles and Pins
The beers I gotta hide...

WOW

cold turkey and no AA???

double absolute awesome!!!!!! i hope you could teach your boys some, um, restraint, they finishing college now, right?

i don't never forget your twins because they born the same day as mine. and dial's grrrl

Sam/Chris - WPA tends to be off (too low early, too high late) - if you are looking at it from the hitter's perspective, because you can't leverage a hitter.

David Ortiz' 9th inning HR isn't any more valuable than ARod's 4th inning HR, because essentially you can't leverage hitters. They come up when they come up. They are each getting 4 or 5 PA in a typical game, etc.. Which is why I think WPA is 100% useless for anything but relief pitchers.

For relief pitchers I think Tango's Leverage Index is much more useful, because relief pitchers can be leveraged and used in those most beneficial situations.

I think this is where the disconnect is coming from.

Repoz, Marty Venker is the name of the Secret Service dj that I was talking about. But I think that he was more into dance music than your crtowd.

David Ortiz' 9th inning HR isn't any more valuable than ARod's 4th inning HR, because essentially you can't leverage hitters.

Something can be valuable even if you can't leverage it. A player who has an unusual ability to hit homeruns in the 9th inning of close games has a ton of extra value for the team, even if doesn't greatly affect the team's decision making.

When Repoz ordered it, I said "don't tell me that, you're on drugs!" And he said "No Craig, I'm not on drugs I'm okay, I was just thinking you know, why don't you get me a Pepsi." And I said, "NO you're on drugs!" But all he wanted was a Pepsi, and I gave it to him.

Nice Suicidal Tendencies reference, directed to the right man for it too.

David Ortiz' 9th inning HR isn't any more valuable than ARod's 4th inning HR, because essentially you can't leverage hitters.

The second part of the statement is true. The first part of the statement is arguable. In terms of direct game impact, the 9th inning HR is of course more valuable - it is more likely to lead directly to a win than the 4th inning HR. The indirect impact does of course need to be considered - without the 4th inning HR, could the 9th inning HR even have happened? - but from what I can tell, distribution and timing do make a difference (albeit not especially large).

-- MWE

Not to devolve into an entire WPA thread (I much more enjoyed talking about SABR), but...

- If the topic is clutch hitting, then I don't see how the "can't leverage hitters" logic applies. The question is whether hitters change their performance in response to game situations, right?

- Further, it doesn't make sense to me to say that some pitchers pitch better in high leverage situations, but that hitters' performance in high LI situations is irrelevant. If one is true, then measuring the other is valid.

- Finally, I know I'm in the minority, but I like having a stat that reflects the progression of the game, the emotional highs and lows of what happened. To me, WPA is a fans' stat and that's why I enjoy it so much.

I'm a pretty light drinker, but it's good to know that I out-drank a few people at SABR. Don't know why, just is.

I like how a thread about getting drunk and playing poker devolved into a serious discussion about sabermetrics.

"A player who has an unusual ability to hit homeruns in the 9th inning of close games has a ton of extra value for the team, even if doesn't greatly affect the team's decision making."

Please point me to such a player. I don't believe this animal exists. Ortiz's entire career is basically right near the 'average' line, except for 2 outlier seasons. Did he forget how to be a clutch God in the other seasons? And even if he did exist, if he hits those HR at the expense of earlier HR - which he must (or how else would we know he had this great ability to summon the inner strength later) he reduces his own chance of being in a high leverage situation (significantly, right?) somewhat dampening the impact of his clutchgodliness.

"Something can be valuable even if you can't leverage it."

Of course it can be valuable. It's as valuable as other similar events that can't be leveraged.

In a 9-7 victory, for example, the grand slam HR in the 4th is exactly as valuable as the grand slam HR in the 9th. I don't see how this is debatable. Without either one you lose 7-5. And since you can't save your best hitters for the 8th and 9th innings of close game, in general, a theoretical player that can only hit HR in the 4th inning isn't any less valuable than an otherwise equal theoretical player that can only hit HR in the 9th.

I say in general, because if you knew the 4th inning player couldn't hit a HR in the 9th (nothing is ever that clear in the real world) and you needed one, you could PH for him in the 9th inning with a lesser overall, but more 'clutch' performer. But no one is pinch hitting for the caliber of player we are discussing in those spots. Regular hitters get a more or less random distribution of ABs throughout a game and season.

If hitters tired like pitchers (meaning they couldn't be used every day) and you could have the best ones take all of their ABs in the late innings of close games, saving the wasted AB of blowouts, then being clutch would matter.

"Further, it doesn't make sense to me to say that some pitchers pitch better in high leverage situations, but that hitters' performance in high LI situations is irrelevant."

I wasn't referring to pitchers pitching better in high leverage situations, I was talking about the fact that good pitchers - just general good pitchers - not 'clutch' pitchers - can be used in high leverage situations, which is why WPA is much more relevant for relievers.

Hitters get 4 or 5 AB in typically average leverage spots. Spot in the batting order is probably the most important way you can leverage a hitter, but isn't the typical hitter's LI in a range of .9 - 1.1 over the course of a season?

And yes, if ARod doesn't hit his HR in the 4th, Ortiz doesn't get a chance in the 9th. That's the main reason why I don't feel WPA is relevant for hitters.

To me, WPA is a fans' stat and that's why I enjoy it so much.

It is fun. But that's where I draw the line in terms of it's usefulness for non-relievers. Especially since it doesn't take into account many, many things. Like park, defense, etc.. It doesn't adjust for quality of teammates, which influences the number hi/low LI situations.

It doesn't adjust for quality of teammates, which influences the number hi/low LI situations.

Not that much, actually. There's very little variation among teams in the % of high LI situations.

-- MWE

I like how a thread about getting drunk and playing poker devolved into a serious discussion about sabermetrics.

Really. "You know you're a nerd when ..."

In terms of direct game impact, the 9th inning HR is of course more valuable - it is more likely to lead directly to a win than the 4th inning HR.

Disagree. The 9th inning HR is more valuable only if the context of the game has changed. In short, the value of the 4th inning HR mediates the value of the 9th inning HR. The 9th inning HR is not more valuable in and of itself. It's one swing, one hit, one run.

Interesting Mike (#147) - I didn't realize that.

Yeah Steve and Studes, I liked talking about the SABR convention too! I blame this whole diversion, like everything else that goes wrong, on Dial. Remember, all things bad can always be traced to the Mets and their fans . . .

"In terms of direct game impact, the 9th inning HR is of course more valuable - it is more likely to lead directly to a win than the 4th inning HR."

"Disagree. The 9th inning HR is more valuable only if the context of the game has changed. In short, the value of the 4th inning HR mediates the value of the 9th inning HR. The 9th inning HR is not more valuable in and of itself. It's one swing, one hit, one run."

That's what I'm getting at. Turn the 4th inning Grand Slam into a GIDP, and the clutch game-winning grand slam down 7-5 in the 9th, becomes an anti-climactic grand slam while down 7-1.

That's what I'm getting at. Turn the 4th inning Grand Slam into a GIDP, and the clutch game-winning grand slam down 7-5 in the 9th, becomes an anti-climactic grand slam while down 7-1.

Exactly. My problem with leverage from way back is exactly this. It attempts to assign value to a variable. I guess it's an interesting exercise to say that "in situation X of this huge matrix this event would change win probability Y" but I'm not sure it's meaningful. The meaning of that ninth inning "high leverage" event is determined by the collected meanings of all of those "low leverage" events that happened before it.

Exactly. My problem with leverage from way back is exactly this. It attempts to assign value to a variable. I guess it's an interesting exercise to say that "in situation X of this huge matrix this event would change win probability Y" but I'm not sure it's meaningful. The meaning of that ninth inning "high leverage" event is determined by the collected meanings of all of those "low leverage" events that happened before it.

Which is why I've always been unimpressed with the value of gaining the platoon advantage by using your 3rd LOOGY in a high-leverage late-inning situation. Dude. Use that roster spot on another lefty bat instead, and gain the platoon advantage instead in literally hundreds of "low leverage" early inning PAs on the offensive side, and it seems to me you come out ahead.

Something for Chris J: Paul Richards pinch-hit for position players in the early innings a *lot* while he was managing Baltimore. The Orioles were in double-digits in that category - and had more than any other team in baseball - in every season from 1956-1961. No team has been in double digits since the Tigers in 1987.

-- MWE

Mike,
Paul Richards was just a bit before my time. Earl Weaver often started his more defense oriented players and then subbed in his more offensive oriented players later in the game if appropriate. Did he inherit that tendency from Richards?

24 hours late on this one, but figured I'd respond anyways.

Maybe Defensive Efficiency falls because teams go into "no doubles" defensive mode and slap hitters are better able to take advantage of their suboptimal positioning to defend against singles.

What were the years that Cralmer used for the study? I think that "no doubles" is recent. At least I don't recall announcers referring to it when I was young. If I had to guess, guarding the lines or drawing the infield in may be a bigger culprit.

I read Weaver's "It's What You Learn After You Know It All That Counts" last year. I don't recall him mentioning Richards despite the fact that he was managing in the Baltimore minor league system since at least 1957.

I wasn't referring to pitchers pitching better in high leverage situations, I was talking about the fact that good pitchers - just general good pitchers - not 'clutch' pitchers - can be used in high leverage situations, which is why WPA is much more relevant for relievers.

I was referring to the comments about Palmer and Glavine. When they didn't relieve.

Back to the convention...

The 9th inning HR is not more valuable in and of itself. It's one swing, one hit, one run.

But it does increase the probability of the batting team winning the game by more than does the 4th inning HR, given the same score differential/out context for both events. That's why I said that it was more directly valuable. Obviously, there are times when without the 4th inning HR you wouldn't get the chance for the 9th inning HR (and I said as much, too), which offsets the direct value of the later event.

You run into all kinds of problems when you try to simulate how performance shifts affect teams; it's difficult at best to account for enough of the variables to get any sort of meaningful result. Consider that:

-- Teams manage their personnel differently when they are ahead than they do when they are behind going into the late innings - if A-Rod hits that 4th-inning HR, it makes it more likely that Mariano Rivera will pitch in the game, and less likely that LaTroy Hawkins will.

-- There is a tendency for teams to be drawn back to the middle; a team that gains an early lead will frequently give some of it back fairly quickly (do pitchers tend to relax after being handed early runs? do the defenses play more conservatively and allow extra baserunner advances? do teams that trail focus more on chipping away rather than playing for crooked numbers? I have no idea).

I agree that getting to a high-leverage late-inning event is largely a function of the aggregation of lower-leverage events that occur prior to that. But I think that that aggregation of events doesn't quite offset the importance of late-game high-leverage performance; that 9th-inning HR would, on balance, lead to more wins than would the 4th-inning HR in the same score differential/out context, even after you account for everything else. I wouldn't use raw WPA swings as a measure - I agree that WPA overstates the impact of late-inning performance - but I think LI goes quite a ways toward putting in-game performance on something close to an appropriate value scale.

-- MWE

Earl Weaver often started his more defense oriented players and then subbed in his more offensive oriented players later in the game if appropriate. Did he inherit that tendency from Richards?

Could be. The Orioles were well up there in the Weaver years, also.

-- MWE

If I had to guess, guarding the lines or drawing the infield in may be a bigger culprit.

I suspect the latter, actually; teams have shown a marked tendency toward playing the infield in more frequently.

-- MWE

I agree that getting to a high-leverage late-inning event is largely a function of the aggregation of lower-leverage events that occur prior to that.

How is it merely "largely" a function of what came before? Isn't it simply the case that the leverage status of every late-inning event is absolutely a function of every event that preceded it?

Isn't it simply the case that the leverage status of every late-inning event is absolutely a function of every event that preceded it?

Not entirely, at least in the way that I look at things, because there are *many* routes that can get you to a certain point. A 1-run deficit, bottom of the 9th, no on and two out, with the cleanup hitter batting can occur in a 1-0 game or an 11-10 game; the leverage of the situation is exactly the same but the path to get there is quite different.

-- MWE

Okay, I'll throw in one more thought. What WPA captures -- and I don't know of any other stat that does this -- is the difference between production in a close game vs. a blowout.

I know some people will have the same complaint, that game differentials are dependent on other events outside the batter's control. But it seems to me that games are discrete units and production should be rated differently based on the context of each game.

I came up with a system that isn't pure WPA, but does give more credit for production in close games, in the second half of this article:

http://www.hardballtimes.com/main/article/long-live-baseball-analysis/

I put it out there, wondering if folks would have the same objections to a system like this.

Not entirely, at least in the way that I look at things, because there are *many* routes that can get you to a certain point. A 1-run deficit, bottom of the 9th, no on and two out, with the cleanup hitter batting can occur in a 1-0 game or an 11-10 game; the leverage of the situation is exactly the same but the path to get there is quite different.

Well ... yeah. I guess the significance of this eludes me. In the 1-0 game, failure of my team's batters up to now has been balanced out by great work by my team's pitchers; and in the 11-10 game it's just the reverse. In either case, this current situation is only as highly leveraged as it is because of everything that's happened before, positively and negatively.

What WPA captures -- and I don't know of any other stat that does this -- is the difference between production in a close game vs. a blowout.

It does, and that's fine as far as it goes. But it seems to me that being on the winning side of a blowout is a good thing, and that doing positive things in the early low-leverage innings that make the game a blowout is every bit as valuable as doing positive things in the late high-leverage innings of a close game.

But it seems to me that being on the winning side of a blowout is a good thing, and that doing positive things in the early low-leverage innings that make the game a blowout is every bit as valuable as doing positive things in the late high-leverage innings of a close game.

Well, I don't agree with that, but the stat I developed doesn't look at when in the game an event occurred. It just weights events according to the final outcome of the game. In close games, home runs have more impact than they do in blowouts. So do strikeouts, for that matter.

I don't agree with that

What part of it don't you agree with?

What part of it don't you agree with?

I think that positive events in close games have more win value than positive events in blowouts.

I SHOT REAGAN!!!

I think that positive events in close games have more win value than positive events in blowouts.

Well, this concept just doesn't penetrate my thick skull.

I'll certainly grant that a home run with two outs in the ninth inning when one's team is already ahead (or behind) by 12 runs is rather valueless. But what I don't understand is how the positive events that made the game a blowout in the first place are any less a contributor to a victory than a clutch hit in the late innings of a close game.

The point is to win as many games as possible, not just as many close games as possible. Doing the things that gain early leads, and pad leads for the precise purpose of ensuring that the game won't be close in the late innings, are every bit as positive as late-inning heroics. They have just as much win value. Blowout wins count equally to close wins in the standings, and the surest path to a win is to have a generous lead by the time the late innings roll around.

I see what you're saying, Steve. To me, the thing we're going for in systems like this is to bridge the gap between production and value. I define value as those events that contribute most to a win.

What I'm proposing is a system that looks at games in retrospect, not in real time. In retrospect, a home run in a close win contributed more to the win than a home run in a blowout. Because I'm looking in retrospect, I don't care when it occurred. As far as I can tell, most people who don't like WPA don't like the "real time" aspect of the system.

Yes, the point is to win games, not just close games. But events in close games have a bigger impact on the outcome.

I think you're proposing something I haven't quite heard anyone propose before: a value metric based on base/out/score situation, but not the inning. So a home run in the ninth of a 2-1 game would be weighed the same as a home run in the second inning of a 2-1 game (regardless of the actual final score of the game). You know, it would be pretty easy to build a matrix like that. It would be interesting to see if the results added up to the exact impact of the win/loss, which is one of the biggest benefits of WPA (WPA=wins-losses). I don't think it would, but it's an interesting idea.

studes, that kind of reminded me of the Victory Important RBI stat that was in one of Bill James's Baseball Abstracts. I think VI-RBI was what steered me towards WPA in the first place.

You're right, GGC. It is very similar to Victory Important RBI. James' ratios were much more severe than mine.

Value is a combination of the context and the participant. I think most disagreements over the utility of WPA arise around how much weight each part of the equation gets. Hitters don't decude what context they hit in but they do execute once put in that context. So to me, it isn't an indictment of the guy that hits a 4th inning homer to say that a different guy that hit a 9th inning homer provided more value.

A bottle of water at store 24 in New York City is not as valuable to me as one would be in the middle of the desert. The bottles are the same, the context is drastically different. I think most disagreements over WPA is how you weight the context/player contributions in things like MVP discussions and the like.

Hitters don't decude what context they hit in but they do execute once put in that context.

Here's my issue with that. Study after study indicate that players don't perform differently in that context than otherwise. Lots of them. So to "reward" a player in that context is bad form to me. WPA is too RBI-ish for me. Everything that creates the 9th inning situ is dependent on his teammates. Player X bats in the 9th and hits a home run. That's all he did - the game state shouldn't be credited to the player *if he isn't creating that state*. He adds one run to his team (or 1.4 runs) - And he should get the same credit whether his team wins or loses - his performance isn't affected by the game state, so his value shouldn't be.

As for the water in the desert. The water in NYC isn't "better" than the water in the desert - the water has nothing to do with the locale.

There is clutch hitting (and clutch water), but there are nearly no clutch hitters (or clutch water). The player (or water) aren't extra valuable - the situation is different.

That paragraph makes you and I both say "Exactly". You think that counts for the player (or water)- I do not.

Study after study indicate that players don't perform differently in that context than otherwise.

Wrong. They DO perform differently. Their performance has a different shape: more walks, fewer hits, lower ISO.

-- MWE

Chris, a thoughful response, thanks. I think we are pretty close in agreement.

But I DO think there is middle ground between attributing all the value to the player in clutch context and not giving him any value above just the single/homer etc.

I've picked up and moved a child many times in my life. I've never picked one up and moved him when he was about to be run over by a car. In the latter circumstance, I deserve more credit, regardless of whether or not I had anything to do with the child/car context arising. I don't deserve credit for doing brain surgery on the child. I have no problem attributing equal value to moving the child away from the car and a doctor doing brain surgey on the child while also acknowledging the brain surgeon had a more difficult/skillful act to perform.

Wrong. They DO perform differently. Their performance has a different shape: more walks, fewer hits, lower ISO.

That's a tremendous nit.

That's a tremendous nit.

Au contraire. If a player keeps his power production up in high-leverage situations, he typically generates more runs for his team. That's why teams on defense are willing to trade walks for decreased power (and why you see the shifts in performance that you do); they'd rather force teams into long-sequence offense, because they believe (with some justification) that it's easier to keep teams off the scoreboard that way.

-- MWE

Excellent points, Mike. It also may help to explain why some players aren't viewed as clutch, particularly if the shape of their production changes in a way that lowers their BA, upping the number of ABs where they didn't drive-in or advance runners.

There is clutch hitting (and clutch water), but there are nearly no clutch hitters (or clutch water). The player (or water) aren't extra valuable - the situation is different.

What are you attempting to measure, the player or the game? If the former, then you're right. More over, even givin Mike's caveat about the different shape of their performance (which I don't think it necessarily a nit), the marginality of the value-add and the unlikeliness of our ever being able to project it into the future with any degree of certainty makes you even more right. But if you're measuring the game, I can see Mike's point of view more clearly.

What are you attempting to measure, the player or the game?

That's a good way to put it. I think you are measuring the game, and then breaking out the impact of the player on that game. That may be trivial wordplay, but I think it's essentially what WPA does.

I agree that WPA doesn't get at the "essence" of a player (or the water) as well as some other stats. But I think it's a unique, interesting, descriptive, retrospective stat.

Everything that creates the 9th inning situ is dependent on his teammates.

Yup, WPA is probably a more legitimate "true" measure of pitching than hitting (Ignoring fielding for a moment, pitchers create their own situations and relief pitchers are expressly brought into situations). For batters, WPA/LI might be a good compromise.

If a player keeps his power production up in high-leverage situations, he typically generates more runs for his team. That's why teams on defense are willing to trade walks for decreased power (and why you see the shifts in performance that you do); they'd rather force teams into long-sequence offense, because they believe (with some justification) that it's easier to keep teams off the scoreboard that way.

Which way is the arrow? Do players take shorter swings trying to get the run in, or does the defense take away those plays? Can you tell?

And how is the RC differential between those two shapes? Is the increase in OBP enough to practically offset any drop in ISO? Is shape important if the RC is approximately the same (given the different baseline)?

What are you attempting to measure, the player or the game?


That's a good way to put it. I think you are measuring the game, and then breaking out the impact of the player on that game.

It didn't occur to me that the way WPA is casually used that it measures the game. I agree with Studes that it is a fun and interesting stat to follow the ebb and flow of a game. However, when we discuss a clutch hitter or an MVP candidate, trying to then measure the player based on WPA (or a derivative) is, er, not the best way to do it.

I agree with Studes that it is a fun and interesting stat to follow the ebb and flow of a game. However, when we discuss a clutch hitter or an MVP candidate, trying to then measure the player based on WPA (or a derivative) is, er, not the best way to do it.

Yes, and yes.

...when we discuss a clutch hitter or an MVP candidate, trying to then measure the player based on WPA (or a derivative) is, er, not the best way to do it.

I agree with the MVP comment, too, but I do think WPA is a good measure of retrospective clutch performance. I also think it can be used as a backup/bench stat for MVP consideration.

Lastly, I think that over an entire 10-year career, WPA is very useful cause its idiosyncrasies tend to even out over a career. Not a perfect stat -- very useful.

but I do think WPA is a good measure of retrospective clutch performance. I also think it can be used as a backup/bench stat for MVP consideration.

It does describe a clutch performance - but I don't want people to claim that the player is thus a clutch hitter. I also agree it can be some of the seasoning to an MVP discussion.

Hey Chris! We agree! BTW, congrats on that nice grab last night.