This isn’t relevant, but could you update the glossary page to address the new pitch type/speed and swing percentages features you’ve added? The page you get when you click “glossary” on the far right of the main tool bar doesn’t have that information. Thanks!
Awesome, i’m hooked. Any idea when ’89-’01 is coming up?
Also, do you plan to allow searches across years? would be nice to see all-time leaders in WPA, Clutch, etc. Also, the ability to combine hitters + pitchers would be nice!
This is awesome, it would be nice to be able to download the whole data base. Thanks!
Comment by Luisarturo Castellanos — May 13, 2008 @ 1:41 pm
Not to be a stick in the mud, but we have to be really careful how we use WPA for things like who was most valuable in, say, 1985.
First of all, WPA if I understand it correctly credits the pitcher for 100% of the “win probability added” on the defensive side of the game, and awards no WPA for plays by rest of the defense. That would substantially overvalue pitchers, such as 1985 Dwight Gooden, in a head-to-head comparison with, say, the most valuable everyday players in the league in 1985 or other years. The Gooden. of ’85 was less dependent on his defense than other pitchers, because of the high K numbers, but still you can’t really give any pitcher 100% credit, or anything close to it, for preventing runs, nor can you give zero credit to everyday players for their performance in the field (other than DHs or below-replacement-level fielders)..
Second, WPA, again as I understand it, has a flaw (for purposes of “most valuable” evaluations) that Bill James identified in Pete Palmer’s old defensive rating system — it is based off an “average” performance standard rather than a replacement level performance standard. An exactly average player (in both a performance and clutch sense) who plays 162 games, though merely average, is way more valuable than a player who plays above average for ten games, then gets hurt and sits the rest of the season. But WPA will rate the second player (who will have a postive WPA) more highly than the full-time palyer (who will have a zero WPA), right? WPA is still very useful for some purposes (say, comparing closers to each other), but it means you really can’t compare modern starting pitchers, whose limited innings necessarily limit the ceiling on their contribution above replacement, to everyday players.based on WPA.
Win Shares has Gooden ’85 as no more valuable than Gary Carter on his own team and slightly less valuable than Wille McGee, Pedro Guerrero, and Tim Raines in the NL in 1985. Win Shares is by no means perfect, but given the inherent limitations, for these purposes, of WPA, I trust Win Shares more in this case. And I’m a Mets fan of forty years standing — Dr. K in 1985 was a god.in my house. But that was not the most valuable non-Bonds season in recent baseball.
birtelcom: It’s true WPA does not take into account defensive performance, but it doesn’t take into account defensive performance for any pitcher, so I think you could safely say that in terms of WPA Gooden was easily the most valuable pitcher in 1985 and it was probably the most valuable performance by any pitcher in the past 34 years. If you want to adjust for replacement player, Gooden is going to pick up an extra win or so because he’s a starting pitcher.
If you want to use a non-leverage adjusted wins (WPA/LI) then Gooden is still #1 in 1985 (minus defense), out of all players. When you adjust for replacement, he’s probably not going to be #1, but it’s going to be close.
You should also understand that pitchers are not necessarily overvalued because they receive all of the WPA credit. Batter/the correct base runner also receives all of the credit. So for any stolen base, a pitcher is getting debited. For any error in the field, the pitcher is getting debited. And also for any good play in the field the pitcher is getting credited. It’s not like the pitcher just gets credit for game saving plays.
I certainly agree that Gooden was the most valuable pitcher in 1985, was one of the most (though not quite the most) valuable players in 1985, and probably had in 1985 the most valuable perfromance by a pitcher in the period 1973 to today. (Win Shares indicates all of that. Win Shares ‘ top pitching seasons 1973 to day are Gooden ’85, Clemens ’97, Jim Palmer 1975 and Ron Guidry 1978). These observations are different than your original one though, that Gooden’s 1985 performance was more valuable than any performance by anybody but a couple of Bonds steroids years, which can only be gathered from applying WPA in an incorrect way.
Most people who study the issue closely agree that something like a fifth to a third of run prevention is attributable to defensive performance behind a pitcher (these are guys who are also hitters), and the remaining majority of run prevention is attributable to the pitcher. To take that significant aspect of added win probability away from the everyday players who are contributing it and give it to the pitchers who are not is overvaluaing pitchers when you compare pitchers and hitters head to head using WPA. The problem is less egregious when comparing pitchers to other pitchers, but even then it limits WPA’s accuracy because pitchers with good fielders behind them will be overvalued compared to pitchers with poor fielders behind them — but that’s true of common pitching stats such as ERA too, so I wouldn’t worry about it that much. The bg issue is you really can’t compare pitchers and hitters using WPA, in the way say Win Shares or WARP try to do, because it’s just noit structured to do so accurately.
And, as I pointed out earlier, it’s not good at properly valuing prolific but merely average performances as compared to short-lived but above-average performances. A guy with one pinch hit walkoff homer but does nothing else in his career will have a higher career WPA than a guy who contributes for ten years at league average levels and average clutch performance (WPA=0). The one at-bat guy did not have a more valuable.career.
I love WPA for what it is: a great way to value player contributions over a single game (the star of the game) or a short post-season series (series MVPs) where players’ extended contributions are irrelvant to the analysis, and the leverage associated with a single play can be definitive. It’s also great for comparing relief pitcher (and pinch hitter) value, because leverage plays such an important role in properly valuing relief pitcher (and pinch hitter) contributions. But for issues like season-long MVPs, evaluation systems such as Win Shares and WARP are far more useful than WPA, because they account for centrally important matters that WPA doesn’t currently incorporate. And the primary thing that WPA adds that Win Shares and WARP don’t, a systematic anlaysis of leverage, will mostly wash out for most everyday players over a long season, as players face the usual variety of high and low leverage situations.
You can use Win Advancement (+WPA) and Loss Advancement (-WPA) in conjunction with WPA to see just how much a batter/pitcher contributed to the wins and losses, which should satisfy your average consistency performer vs “one hit wonder”, scenario.
Anyway, I’m not going to disagree with you that Gooden’s season wasn’t better than any non-Bonds season and I was merely making an observation I thought was interesting in my original post. However, I do think you’re undervaluing the uses of WPA/Win Expectancy stats. Someday we will get around to looking at Win Expectancy using batted ball location data….