Well…this should tide over all those Win Probability Addicts who are tied off out there.
WPA differs from WAR in that WAR attempts to assess the overall value a player has provided without reference to game situation. Basically, WAR assumes that a player has little or control over exactly when he gets hits or outs, and thus attempts to assess true skill level, while factoring out random luck. WPA, on the other hand, doesn’t care about true skill level at all. It only cares about how much a player actually helped his team win, based on context. In other words, how “clutch” players were.
Which is why WPA is so perfect for the old school writers and baseball men. Because it is measuring *exactly* what people always thought they were trying to measure with Wins and RBI: how much you helped your team win.
So if we think about the traditional main criteria used by old-school baseball writers to award the two major awards, MVP and Cy Young, which are of course Wins (followed, to a lesser extent, by ERA) for pitchers and RBI for batters (followed, to a lesser extent, by homers), we see that WPA is actually the best way to determine these awards, if these people actually want to measure what they say they want to measure.
...But the American League is a different story, as WPA shows how Wins and RBI fail to tell the whole story of who happened to be the most clutch. Zack Greinke is first overall in WPA, despite being behind the pack in Wins, and Johnny Damon is your leader among position players in WPA.
Indeed, WPA suggests that Zack Greinke was actually the most valuable player in the American League in this particular season, and probably should be the MVP, but if we adhere to the traditional rules that pitchers should not be MVPs and the MVP has to come from a playoff-bound team, than by all rights Johnny Damon ought to be your American League MVP frontrunner.
Reader Comments and Retorts
Go to end of page
Statements posted here are those of our readers and do not represent the BaseballThinkFactory. Names are provided by the poster and are not verified. We ask that posters follow our submission policy. Please report any inappropriate comments.
And 5 of the last 16 MVPs did not come from playoff teams.
It's funny to see someone combine such traditional fedora-wearing sportswriter thinking with newfangled sabermetriciana like WPA. Anyway, if WPA thinks Johnny Damon is the MVP, that says more about WPA than it says about Johnny Damon.
I actually like WPA for MVP discussion, though I don't think it's anywhere near an end of discussion thing, but I think Johnny Damon is exactly what you're looking for.
* Well, Greinke, but that ain't happening. He's not even going to win the Cy Young.
We go through this every year, some yahoo uses WPA to "decide" who should be MVP, and on some other thread someone else claims that no one uses WPA that way.
To me WPA is like RBI and pitcher W-L records, it is a team and happenstance dependent stat that has some value, but the misuse of the stat by its adherents is so pervasive it not only negates the stats' value but actually gives them negative value. IOW in practice they only serve to confuse the issues
Just to be sure, in adding up WPA numbers, does it matter if the player's team actually wins or loses the game? IOW would Damon's WPA total get inflated just by being on a (so far) 90 win team?
EDIT: Or if not, would his numbers be inflated by the number of clutch opportunites he's had, regardless of whether or not his team won the game? If so, wouldn't a rate stat version of this metric be preferable, subject to a minimum number of plate appearances?
- does not factor position
- does not factor defense
It's useful as one metric for evaluating who is "most valuable", but not the only one.
How does this joke continue to be perpetuated? Does someone have some kind of financial stake in its success? I'd rather look at productive outs!
Mauer will win. Rather comfortably, much like last year's he's-so-overqualfied-he's-going-to-candidate Pujols did, the idiotic ramblings of attention-seeking and non-voting bloggersa and columnists notwithstanding.
No, WPA are accumulated at the time of the PA, i.e., before you know the final score of the game. At the team level, they'll end up adding up to your team's (W-L) but leading off the game with a solo home run produces the same WPA if your team goes on to win 1-0, goes on to win 13-2, or goes on to lose 26-1.
You also get less credit for hitting a home run in the first inning than for hitting a single in the 9th even if the score of both games is 1-0. What a silly stat.
But I'll say this for it: If nothing else, it confirms my thought that Damon's been the Yankees' best "clutch" hitter over the course of the year. Nice to know that it wasn't just a case of selective memory.
It's not a good value stat, for the reasons elaborated here and in every other thread where people use WPA as a simple value stat. WPA/LI is a pretty interesting value stat.
EDIT: Damon's 5th in the AL in WPA/LI. That's pretty interesting. Once you weight by LI, you get rid of the Sac Fly in the 9th vs. HR in the 1st problem, as I understand things, though I'm sure the stat has other issues (such as the 1st inning homer in a 1-0 game vs 1st inning homer in a 13-2 game). And, of course, it's an offense stat with no accounting for defense.
EDIT2: Leading the league in WPA/LI? Joe Mauer. There ya go.
That said, statements like this drive me crazy: "WPA, on the other hand, doesn’t care about true skill level at all. It only cares about how much a player actually helped his team win, based on context." The idea that WPA tells us a player's "actual" or "real" contribution to the team is made all the time, and it's simply wrong. It tells us how he impacted his team's likelihood of winning at different moments in time, which is quite different. Since WPA doesn't "know" what happened next, it can't possibly tell us the real, post-hoc value of events.
Simple example: two teammates hit a 2B and 1B in the seventh inning to score one run, when team was down one. Here is WPA for two versions of this:
Out -.048
Out -.037
2B +.054
1B +.218
Out -.040
Now another version:
2B +.134
1B +.158
Out -.055
Out -.050
Out -.040
The players did the exact same things, with the same net impact on their team (tying the game from one run deficit). Yet in one case the double is valued 2.5x as much. And note that the double is never worth as much as the single, even though this player was just as "clutch" in terms of getting his hit at the right time (right before the other player's single). Since WPA only knows what came before, not after, it rewards clutch RBI performance more than clutch run scoring performance. (Come to think of it, maybe it is a good fit for MVP voters!). But the only way in which the single is different is psychological -- the hitter knew a runner was on base, while the double hitter couldn't know what the next hitter would do. That may be an important difference in the "story," but has nothing to do with the ultimate value of their hits.
You also can get less credit in a game for hitting 2 solo HR rather than 1.
I understand how it doesn't factor position. Does the defense part fall exclusively to the pitcher. Does a fielder not get dinged for errors? Can an OF drop three straight popups to lose the game in the 9th and have all that go on the pitcher?
it's useful at the point in the discussion where you start to list RISP as a stat, it is not useful as the first, second, third or even tenth point in the discussion when it comes to mvp arguments. I would prefer to use wpa/l1 over risp and even wpa over risp numbers, to show clutchitude, but it's not remotely close to answering who is the mvp, it comes to after figuring out that the players have been almost equal based upon everything else, and you don't want to make a tie breaker vote based upon a teams record, then you might use wpa.
If I puked in a fountain pen and mailed it to the monkey house I'd get a better MVP ballot than I would using WPA.
How does this joke continue to be perpetuated? Does someone have some kind of financial stake in its success? I'd rather look at productive outs!
I wouldn't go that far, but it's a pretty accurate statement, I would never use productive outs in an mvp discussion, I can see WPA late into the discussion.
Melvin Mora?
Negative.
WPA/LI has no issues like that, whatsoever. If someone wants to say that there are issues, then please say what those issues are, and I'll refute them.
What WPA/LI does, and it is the ONLY stat in existence that does this, is to make sure that the walk-off HR and the walk-off walk have the exact same value (roughly +.06 wins), and does so without the inflationary aspect of the game importance (likely get say a +.60 wins using WPA).
The only issue you can have with WPA/LI is that I have not explained it well. It is otherwise the perfect stat (if you believe that a batter/pitcher will approach the base/out and inning/score situation differently from random).
***
As for decring WPA or WPA/LI for not accounting for defense, well, neither does OBP! And OBP weights a walk and HR as a "1" in all situations. Down with OBP.
Right, WPA is a "story stat" (as studes has coined it). It quantifies how you feel at the time the event takes place. That is what it is.
If someone wants to use a hammer to nail in a screw, what do you want us to do about that?
On WPA; Or, why
Johnny Damon is your 2009 AL MVPWPA is a junk stat, and is fairly useless unless you are talking about a reliever, who can actually be leveraged.Fixed.
Thanks for clearing that up for us, Tom.
Since you asked, it's all there.
UPDATE: this was the original, less technical article I wrote. As I said, this is a tough thing to try to explain.
You are wrong Joe. Just because you haven't found a use for it doesn't mean it is useless.
The issue is that in a 5-4 win, ARod's proverbial 3rd inning 2-run HR is as important as David Ortiz's proverbial 9th inning 2-run HR. Without ARod's HR, Ortiz's HR is a footnote on a 4-3 loss.
If you could leverage Ortiz to bat 4x a game in high-leverage situations, WPA would have value for hitters, but you can't, so it doesn't. That's why it is reasonable for relief pitchers and pinch-hitters as a metric.
But for hitters it's a junk stat, that is highly dependent on teammates as well.
I think you made a mistake somewhere in that analogy... or I'm not following it at all.
I haven't read carefully through your links, so my apologies if this is well-explained and I just haven't seen it. But I don't understand why you would want a stat that does this. Isn't the only reason why "the walk-off HR and the walk-off walk have the exact same value" precisely because of "the inflationary aspect of the game importance"?
It's not you CFB.
In no way is your scenario true for WPA/LI. Indeed, the two events are likely very roughly equal.
If you want to say that WPA doesn't address your particular issue, then that's fine. What WPA says, and the only thing it says, is that: GIVEN the future outcomes are unknown, how much has the chance of the team winning changed following this event. That is it. That's all it is. It quantified your feelings at the time you felt it.
If you think this is useless, to you, that's fine. But, I find it useful for analytical purposes to see who was involved in the biggest plays (at the time they happened).
It has use, like when Sean Forman shows the 'game-changing' hits from the night before on B-R.
But as a value metric, it might not be useless - in general the better players will have higher WPA. But it's not a good metric for rating players if you want to know who the best, or most valuable players are. It's an eyeball metric that is fun to look at, and nothing more.
If anyone came in here declaring that the league leader in OBP was necessarily the MVP, he'd get hooted down, too.
This is news to me. If the metric treats those two events as equal, then I definitely could see some value in it.
No. I explained it in the article I linked, and I'll repost it for you:
Basically, while a walk is normally worth +.03 wins and a HR is worth +.13 wins, in this particular situation they are both worth +.06 wins.
See? There is no inflationary aspect here. They both get the same value, because they are worth the same thing.
Assume ARod and Ortiz are teammates in the example. I was just using the posterboys for the stat . . .
Neither HR counts. They're both juicers.
Yes, a perfect example. WPA tells you the events that caused the biggest cheers (or moans). It goes farther in that it also tells you WHO was involved. And it goes farther still by TRACKING who was involved over the season. And it goes even farther by compiling leaderboards.
If Johnny Damon is the leader in WPA in the AL, that simply means that he was involved in a ton of big plays. Whether he was lucky to be there is irrelevant to WPA. It simply tells us that he was there when great things happened.
You can call that useless if you want.
I totally agree that the league leader in WPA does not an MVP make. I'm with you there.
But, in no way that because someone says that does it make WPA or OBP a junk stat. It's a junk application.
Thanks, Tango. I think I see what you're trying to do there.
This may be true.
But I've read his explanation half a dozen times now and I still don't understand his stat. I don't even understand what he thinks it measures, much less how it can be the "perfect" stat. And yes, I've read TFA linked to here.
Given the utter bafflement by a LOT of people about this thing, can you cut out the tone of righteous indignation every time it comes up?
Read the link in post 30.
For all the people who think that a strikeout with a runner on 3B and less than 2 outs is an extremely bad event, then this stat is for you. If you think that a K is like any other out in this case (even a SF!), then this stat is NOT for you.
as to the second half,(the wpa portion that isn't the wpa that we are used to ridiculing, which has hampered me into understanding that aspect. I think I understand it to an extent, and I guess it's good to help people who want to say Arod hits homeruns only in garbage time, still not convinced that any game state type of stats is a good idea when evaluating a players season, but that is probably my perception propblem.
ok, that is what I thought you were trying to get at, but wanted to clarify.
A simple NL MVP ballot, compiled by adding WPA to Fangraphs' position adjustments at an exchange of 10 runs per win:
Pujols
Fielder
Utley
Howard
H. Ramirez
Votto
Ethier
Sandoval
Wright
Reynolds
Not a ballot I'd submit, but it wouldn't surprise me if it ends up being more reasonable than the average actual ballot.
And what about the people who think this stats says who should be the MVP? That is what gets most people's dander up, and I should think it should get yours as well. The "It's only a junk stat" comments are mainly an understandably overarching response to the "Damon is the MVP because he's leading in WPA" comments.
ok that is taken care of in the wOBA stat though, when you start to deal with game states though is where you start to lose me to be honest. I'm pretty sure I understand what wOBA does(now) but not 100% sold on the need to figure out game states, I mean wOBA does a good job of saying what a batter does in a particular situation and how it helps/hurts the team in trying to improve their ability to score runs. Those are somewhat controllable during the inning, but when you factor in the game states, you start to mess with stuff more out of the players control during that particular plate appearance. There are some advantages as you point out about the bases loaded walk off hit, but for the most part, when I'm evaluating a player I'm more concerned with how he helps the team, independent of the teams wins/losses.
You are in the huge majority. I take it as a failing of mine that I can't make the explanation better than I have.
My tone is proportional to everyone else's.
But once you start weighing late innings of close games more importantly than early innings of games that eventually become close in the late innings, that's where I jump off the train.
IIRC, the Mills Brothers factored in errors. Not sure if Studes does. I grokked Tom's explanation of WPA/LI last time this came up. If I have time today, I'll see if I can read it and explain it.
As I said, it's a junk application. That doesn't make it a junk stat.
If people were to say: "Yet another article about the misuse of WPA", then I'd be in agreement with those people. Instead, it's always the same "Trying to use a hammer on a screw? Who's the crazy guy championing the hammer?"
You'd think we'd all be on board here, but I seem to have the same discussion with the same people all the time.
your tone has been fine on this thread, not sure where he is getting negatives in your tone.
You are in the huge majority. I take it as a failing of mine that I can't make the explanation better than I have.
the second link you gave does a good job of explaining wOBA, but it gives short shrift to the game states explanation in my opinion. There is almost no explanation on the leverage index portion of that link. You give one extreme example at the end of the game that helps, but there is no (or maybe a link to..) leverage index explanation.
I can see other issues on clarity of explanation to make wOBA easier to understand, but at least you have all the information there. Then at the end you say "All we need to do therefore is take the win impact of each event and divide it by the leverage index" yet don't give the chart or link to leverage index. (unless I missed it upthread)
I still would have included about three examples of how leverage index figures into the final numbers(three examples is always a good rule of thumb) so that people can grasp what impact it has, and whether it's better than WPA.
WPA/LI doesn't by design. The dividing by LI normalises each plate appearance to carry exactly the same weight.
attempted explanation: WPA gives the change in win expectancy incorporating all game state information - this adjusts production valuation from normal to that set of circumstances. This "correctly" rewards such things as hitting home runs with bases loaded compared with solo shots, striking with the bases empty relative to having a man on third etc. This type of thing is maintained in WPA/LI. The criticism of WPA seen above is that it weights plate appearances in games differently based on how important that plate appearance is known to be at that point in time (whether this good or bad depends on your usage/point of view). THIS IS REMOVED by dividing by the LI.
In a nutshell: WPA adjusts the way shape of production is valued for game state and multiplies by percived importance.
WPA/LI adjusts shape of production but weighs all plate appearances equally.
BTW, did anyone ever answer #12? Where can I see the leaderboards for this . . .
And this has ZERO to do with WPA/LI. Indeed, WPA/LI goes out of its way to ensure that each PA counts as "1".
WPA goes out of its way to ensure that some PA count as "10" and others as "0.01".
Basically, you are not on the train you think you are on.
You were so close. WPA/LI has NOTHING at all with whether a team wins or loses, and whether that players advances a team to winning in any way other than what we normally think (HR good, out bad).
It simply reweights each event up and down a little. Like, in this case you may say the single is worth a bit more than normal and a double is worth a bit less than normal. OVERALL, the average positive event is identical in every game situation. Sometimes a walk is worth +.02 wins, other times it's +.06 wins, but it's all centered around its average of +.03 wins.
It's simply a realization, like with the K, that each event has a different RELATIVE impact to all other events. But, the average impact is identical.
WPA/LI adjusts shape of production but weighs all plate appearances equally.
clarification: it doesn't care one whit about score of the game or not? whether you win or lose?
You are not a big Fangraphs user I see.
that is the clarification I was looking for on WPA/LI. (I've actually been told this multiple times but then reading the explanation it seems to say that there is something to wins/losses) I like a system that uses base states, that makes perfect sense, and when Wins probability added came out, it mirrored something that I had thought of in my head(as a concept) so initially I liked it, then weird numbers/names started popping up on the leaderboards and the flaws that didn't occur to me originally in that concept showed their ugly heads, and it really turned me off to the concept of using a stat based upon team win/loss concept. I'm still fine with base states, even though I can see the arguments against it(which is a muffled version of why should a players teamates affect his value argument--as a reminder, it's a team sport,--it's why I much prefer era over fip for debating cy young also.)
It only cares about the score/inning if it impacts the value of the events RELATIVE TO EACH OTHER. So, bases loaded 2 outs, tied game bottom of the 9th, a walk and HR are identical. So, it cares about it that way. But that's the same kind of caring with the K with a runner on 3B and less than 2 outs, or a GB with a runner on 1B and less than 2 outs, etc. It's just more nuanced.
The important takeaway is that while a walk is generally +.03 wins, it is not ALWAYS +.03 wins. It hovers around there.
basically the situation you are talking about isabout the only time it cares? 7th inning no outs, bases loaded down by 3 runs homerun is worth the same if it's 3rd inning no outs up by 3 runs? and of course is the value the same between 7th inning no outs bases loaded homerun vs 7th inning two outs bases loaded homerun?(my assumption on that last one is no because of the outs but just checking)
If so, it seems to me you would get the same result from the RE table, without ever using WPA or LI. Maybe it would be better to describe it that way, as "run probability added" or just "Runs Added," so those disgusted by WPA can better see its value.
depends on who was talking. I was making a mistake earlier in the thread and thinking wOBA definition was WPA/LI. Basically if I said wOBA prior to post 64, I was actually mostly meaning WPA/LI. they are different stats wOBA doesn't take situation/base states into account(I thought it did earlier in the thread)
wOBA is just wOBA = (0.7 x BB + 0.9 x 1B + 1.3 x XBH + 2.0 x HR) / PA anything else was me adding something that wasn't there.
You can create a game-state-specific wOBA equation, such that wOBA gives a weight of 1 for a walk, hit, and HR in the "walk-off" scenario I described.
The common wOBA equation (.7 for a walk, .9 for a single, 2.0 for a HR) is basically the overall averaged version of all the possible game-state-specific equations. It is scaled like OBP.
WPA/LI is the game-state-specific wOBA equation, but denominated in wins.
EDIT: Thanks, Tom. You were posting that answer as I was writing this one...
I'd have to run that exact scenario. It would probably be "close".
The important takeaway though is that if a HR is less valuable in one situation, then the walk and singles are worth more in that situation. The overall positive event in the above two situations (in any two situations actually) is identical.
This is unlike WPA where if a HR is worth more, then it's likely that all the other events are also worth more. And the average positive event in one situation had no bearing on another situation.
***
For those asking for LI, there's a three-part article, along with the LI chart.
Close. First off, the stat you mention is on Fangraphs and it's called RE24.
The difference is that a bases empty situation doesn't give you much chance to add runs, while a bases loaded one does. So, a guy who has a disproportionate number of BL-to-BE situations will come out better than one who doesn't even if they have identical hitting stats.
This is what I'm talking about in the article, which DOES describe RE24:
So, RE24 is halfway between WPA/LI and WPA.
looking at Toms original formula from the book to the newer formula it looked to me like he changed his mind on that the formula I put up in post 71 doesn't remove IBB (unless he made a mistake with the formula when he wrote it. he calls NIBB--non intentional base on balls---in the original formula, uses BB in the newer formula)
In wOBA, it is ignored in the numerator and denominator (like a sac bunt).
In WPA/LI, it gets the value of a regular walk (and since the IBB occurs in disproportionate game states, it comes out to having a value that is half that of a regular walk).
In WPA, it gets the leveraged-value of a regular walk.
My biggest problem with my experience with WPA is the lack of fielder credit. All +/- goes to the batter and the pitcher which makes sense for the 3 true outcomes but not for a base hit. Failure to field a ball isn't only the pitcher's fault or the batter's credit.
WPA/LI appears to be expressed in batting wins above average. And it treats a walk and homerun the same if the bases are load in a tie game in the bottom of the ninth (to name one difference.)
not a big fan of that concept personally. Heck Albert has scored multiple runs last month alone because the opposition thought it was better to intentionally walk him and face Holliday in a close game.
and of course the intentional walk that happens after a wild pitch advances a runner in the at bat doesn't really do the batter justice either.
Then you must have a problem with a pitcher's OBP and SLG allowed then.
Seriously, the stat is what it is, and you can't ask more from it than what you ask of the staples.
But WPA/LI DOES count the IBB, and counts it properly.
Hmmm. I think of leverage as having two components: the base/out situation, and the inning/score situation. WPA takes both into account. It also seems useful to me to have a stat that ignores inning/score but does take base/out into account (i.e. how much the hits/walks contributed to scoring). It sounds like that is what RE24 does. Right?
Now, can you explain in words how WPA/LI differs from RE24. What does it include that RE24 doesn't? What does it exclude that RE24 includes? And why might one prefer WPA/LI? (Sorry if I'm being obtuse.)
Seriously, the stat is what it is, and you can't ask more from it than what you ask of the staples.
the thing is that this article pretty much says that Damon should be the MVP because of his WPA, acting as if it's a be all end all stat and yet it isn't. As you said earlier it's a junk application, the problem is that fans of WPA don't think it's a junk application and will sometimes begin and end with the mvp argument based upon WPA when they get their guy that they want to champion.
But WPA/LI DOES count the IBB, and counts it properly.
yes, I'm 85% sold on WPA/LI still debating on whether I agree that a grandslam in the bottom of the 9th of a tie game should be considered equal to a bases loaded walk in the same situation. (I understand, just internally debating if I agree)
I can debate it for you. If you think the batter/pitcher change the approach to that particular game state such that they both realize that a walk is a HR, then, yes, you have to consider them equal (the players do).
If you think they don't change their approach, OR, you think that hitting a HR in that situation is as indicative of talent as hitting a HR in any other situation, and drawing a walk in that situation is as indicative of talent in some random situation, then NO, you do NOT want WPA/LI.
RE24 counts a PA with bases loaded (about 2.8) more than it counts a PA with bases empty (about 0.7).
WPA/LI counts a PA identically across all game states (they all count as "1").
You can do RE24/LI to fix that problem.
So, the question is whether you prefer WPA/LI to RE24/LI. And the answer depends entirely on whether you think batters/pitchers change their approach only on the base/out states, or they also consider the inning/score.
If you don't believe they do either, then you want a straight linear weights (or wOBA or EqA).
***
We all accept they change their approach by count. I'm of the opinion they change it all the way.
I'll try and Tom can correct me if I'm wrong: RE24 is expressed in runs while WPA/LI is expressed in wins. Also RE24 takes into account the base-out state but doesn't take into account the inning and score. Now, the way WPA/LI differs from WPA is that it puts everyone on a more level playing field, so to speak.
Yes, Sandy Koufax found this out as Willie Davis lost the last game Sandy ever pitched.
OK, RE24 takes into account the base-out state but not the inning and score.
What does WPA/LI take into account that wOBA doesn't? Or is it just wOBA converted into wins?
I meant this as WPA doesn't factor position and defense the way a stat like WAR does. WPA, for a hitter atleast, is all about offense regardless of position and doesn't factor the player's defense contribution at all. Hence my point that WPA alone should not determine the MVP.
Every team needs a shortstop. Every team needs a first baseman. If the unadjusted numbers of a shortstop and a first baseman are the same, the shortstop was more valuable, because the pool of players you can select from to replace him are not as good as they are if you're replacing the first baseman.
This is precisely what I was trying to say in my first sentence. "I meant this as WPA doesn't factor position and defense the way a stat like WAR does."
In wOBA, a HR in this situation has a linear weights value of about 1.4. That's the general value of a home run averaged across all situations.
In WPA/LI, its linear weights value is 1 - a HR has the same value as a single or a walk, all positive events in this limit scenario have the same linear weights value.
What WPA/LI does is look at how HRs and singles and Ks and other batting events have different relative values based on base-out and game state. By dividing by LI, the formula does not give extra credit (or demerit) to players who happen to have a lot of highly-leveraged PA. (As would happen in WPA, or in Run Expectancy Added, for players who have a lot of PA in situations with more highly leveraged base-out states.) Rather, it weights batting events based on their relative value in the particular game state situation.
What WPA/LI adds over Run Expectancy / Expected Run Value of Base-Out State is the consideration of game state, which matters in something like the bases loaded, bottom ninth situation above. RE/ERV would give that HR a weight of about 1.15, whereas WPA/LI weights it just at 1.
EDIT: Two clarifications. First, the reason that the lwts value of a HR in the bases loaded, two out, bottom 9th, tie game situation is 1 is that all positive batting events (HR. single, walk, whatever) have the same game-state result - a win. They have, thus, the same relative value.
Second, a stat like WPA or Run Expectancy Added would give a HR in the limit state a much higher value. For run expectancy added, the linear weights value would be closer to 3.3, to account for the large number of extra runs added by hitting a home run in that precise situation. In WPA, the value would be about .5 wins (maybe a linear weights value of about 5, if converted to runs), because hitting a home run won the game. Stats like WPA/LI and Run Expectancy / Expected Run Value cut down on the lwts value of these events significantly, in order not to have the happenstance of batting in lots of or very few important situations be the driver of a players' statline.
EDIT2: Jeez, post 101 is like the highlight of my btf posting career. Hopefully these edits don't ruin it...
Well, the wOBA that is published is a pure linear weights equation.
WPA/LI is the game-state-specific wOBA. For example, the walk in the basic wOBA is always 0.7. In WPA/LI, the walk will be anything between 0.3 and 1.0, depending on how much win impact the walk has in that particular game-state. For example, in your traditional IBB situations, the walk value will be around 0.3. In your walk-off scenarios, the walk will be 1.0. The "0.7" that the standard wOBA equation uses is just an overall average.
Right. You need offense, defense, position, park, and opponent.
WPA is hardly the answer, either. Neither is VORP, or batting average, or run expectancy added. My preference is to have a lot of different ways of talking about ballplayers and what they do - we're never going to stumble upon the perfect number, so it's best to have a bunch of things that serve different purposes, and use all of them with an understanding of their uses and limitations.
You must be Registered and Logged In to post comments.
<< Back to main