Interesting. What might be better though would be a modified form of WPA that is based not on the chance of winning any individual game, but winning the series as a whole. Winning game 4 when you're already down 3-0 isn't nearly as valuable as winning game 7 (or even game 1 or 2), and just accumulating WPA won't reflect that.

Worst. Stat. Ever.

I find that quite interesting. I was looking at players with an extreme WPA in just a few ABs, and came across Cliff Bolton. He had just two World Series at bats in his career, and they both ended games. In the 1933 World Series, he pinch hit with two out in the top of the ninth with the Senators down 6-1, and grounded out to the pitcher. Then, in the bottom of the 11th in game 4, with the bases loaded and one out and the Senators down 2-1, he grounded into a 6-4-3 double play.

Interesting. What might be better though would be a modified form of WPA that is based not on the chance of winning any individual game, but winning the series as a whole. Winning game 4 when you're already down 3-0 isn't nearly as valuable as winning game 7 (or even game 1 or 2), and just accumulating WPA won't reflect that.

Tony Womack had a really good WS WPA if you calculate it like that.

I keep waiting for somebody to develop a Standard Deviations of Win Probability Added over Average player stat, and it never happens.

I understand why people like WPA, but it has the same flaw as RBI's. It depends heavily on opportunity and doesn't strike me as measuring ability or being all that predictive compared to other stats (I say without actually studying WPA to see if it is particularly predictive.)

I was looking at players with an extreme WPA in just a few ABs,

highest WPA per PA (min 5 PAs):

Del Unser 0.098
Dusty Rhodes 0.094
Hal Smith 0.084
Sam Crawford 0.064
Larry Doyle 0.061
Harry Hooper 0.059
Fred Luderus 0.057
George Davis 0.057
Les Mann 0.057
Solly Hofman 0.056
Joe Tinker 0.054
Duffy Lewis 0.050
Kirk Gibson 0.049
Ty Cobb 0.049
Frank Chance 0.048
Claude Rossman 0.046
Willie Aikens 0.045
Buck Herzog 0.045
Harry Steinfeldt0.045

Del Unser was 3 for 6 with 2 doubles and 2 RBIs (and we all know about Dusty Rhodes and Hal Smith)

well, "we" (greater we) all know about Dusty Rhodes. Very few know that Hal Smith had the biggest WS WPA plate appearance in the history of the game. If all sportswrtiers had written much about Hal's blow, Maz still would not be in the Hall.

I understand why people like WPA, but it has the same flaw as RBI's. It depends heavily on opportunity and doesn't strike me as measuring ability or being all that predictive compared to other stats (I say without actually studying WPA to see if it is particularly predictive.)

I may be in the minority of people who find a use for WPA, but for that use predictability is entirely irrelevant.

I like using it for situations like this. Every post-season I usually follow a couple series using WPA just to catalogue the big moments, and put some kind of rough number on the influential players.

I guess it's just a little bit of mindless fun...but since when does being a baseball fan exclude that?

I may be in the minority of people who find a use for WPA, but for that use predictability is entirely irrelevant.

Just because it doesn't have predictive value, doesn't mean it is useless. And it certainly isn't mindless. For analyzing something which already happened, it's incredibly useful (although you have to take it with a grain of salt or three).

Just because it doesn't have predictive value, doesn't mean it is useless. And it certainly isn't mindless. For analyzing something which already happened, it's incredibly useful (although you have to take it with a grain of salt or three).

I agree with the sentiment, disagree with using the word incredibly there. It's moderately useful, maybe, but incredibly implies it actually has some type of real value. It doesn't. It's primary benefit is as a nice little toy to analyze a single game, anything more and you are attaching a value to the stat that it doesn't really rate. If you look at seasonal WPA and assign value to it, it's no more different than assigning value to rbi.



As to the article, nice article, surprised I didn't see Lou Brocks name higher up there, wonder how stolen bases were calculated in this exercise.Not that it matters looking at the game logs seems most of his games were blowouts one way or the other.

Hmm, on second thought I shouldn't have said 'incredibly'. You're right on that account. But it's useful.

Mike Stanton? I would never have thought he'd be top twelve in an all-time list of anything. That's great.

Elizabeth Cady Stanton
Edwin Stanton
Andrew Stanton
Mike Stanton
Other Mike Stanton

I think his Top 12 slot is pretty secure.

I guess it's just a little bit of mindless fun...but since when does being a baseball fan exclude that?

Hence my "I understand why people like it". I just don't think it's the a particularly relevant stat.

I am serious about wanting to see somebody try to do a standard deviations of WPA over average stat; I feel like that could be a useful stat. I could be horribly wrong, however.

#10 I think it's a very useful sanity check on certain forms of narrative. As in Mark Lemke, Clutch Lord.

And WPA does do a nice job when you're looking at relief pitchers.

And WPA does do a nice job when you're looking at relief pitchers

Provided you don't compare them to other players(other relievers is ok, other players not so much), then yes I agree on that. I'm still waiting for any "uber" stat that I feel comfortable comparing across positions. I'm fine with war for everyone except catchers, firstbaseman and pitchers(although I've gotten more comfortable lately with all three just still not sold) and I don't hate WPA as much as it sometimes seems, it's just that it is a niche stat and is perfectly useful when used in it's proper niche with tons of disclaimers, it's when people try to take it out of that niche that it bothers me. For the most part it's not any better than RBI with the exception that it effectively has a rate component built into it.

Elizabeth Cady Stanton
Edwin Stanton
Andrew Stanton
Mike Stanton
Other Mike Stanton

I think his Top 12 slot is pretty secure.

You forgot Arch Stanton.

What I want to know is, who are the all-time WORST players in the World Series according to WPA?

There's a spreadsheet, Dag. Marv Owen was the biggest goat. WRT pitchers, Eck, BK Kim, and Mitch Williams are down there.

You forgot Arch Stanton.

Overrated. Dig past the surface and there's nothing of worth there.

What I want to know is, who are the all-time WORST players in the World Series according to WPA?

Aaron Boone -0.700
Phil Rizzuto -0.700
Mariano Duncan -0.710
Red Ruffing -0.710
Glenn Wright -0.720
George Earnshaw -0.730
Kent Hrbek -0.730
Walter Johnson -0.730
Andy Pafko -0.750
High Pockets Kelly -0.750
Ozzie Smith -0.750
Steve Garvey -0.750
Javy Lopez -0.760
Morgan Ensberg -0.760
Graig Nettles -0.780
Jorge Posada -0.780
Travis Jackson -0.780
Bill Russell -0.790
Ossie Bluege -0.790
Davey Johnson -0.800
Jimmie Wilson -0.800
Johnny Hopp -0.800
Denis Menke -0.810
Bernie Williams -0.850
Frank White -0.850
Shane Mack -0.880
Orlando Cepeda -0.900
Max Bishop -0.940
Dave Bancroft -0.980
Jim Gilliam -0.990
Ron Gant -0.990
Earl Smith -1.100
Tony Kubek -1.120
Chick Hafey -1.170
Marv Owen -1.260

so, did Gant get penalized for Hrbek pushing him off the base? (and Kubek for the throat incident)?

Interesting. What might be better though would be a modified form of WPA that is based not on the chance of winning any individual game, but winning the series as a whole. Winning game 4 when you're already down 3-0 isn't nearly as valuable as winning game 7 (or even game 1 or 2), and just accumulating WPA won't reflect that.

I actually have this... haven't completely finished checking the lists for duplicate names, and don't have a forum to publish it in. And the data set is the entire postseason, not just the World Series (with LCSs weighted half as strongly, LDSs 1/4). Anyway, here are the lists:

Top 10 hitters:
Mickey Mantle +.826 championships added
Pete Rose +.785
Lou Gehrig +.668
Hal Smith +.655
Reggie Jackson +.617
Dwight Evans +.584
Yogi Berra +.580
Charlie Keller +.543
Tris Speaker +.540
Max Carey +.513

Bottom 10 hitters:
Marv Owen -.486
Graig Nettles -.524
Kent Hrbek -.528
Denis Menke -.535
Art Fletcher -.539
Jeff Blauser -.563
Reggie Sanders -.564
Jorge Posada -.575
Cesar Geronimo -.585
Ron Gant -.743

Bottom 10 pitchers:
Byung-Hyun Kim -.359
Al Downing -.368
Bob Friend -.379
Jack Bentley -.409
Bob Stanley -.425
Calvin Schiraldi -.444
Don Newcombe -.504
Jim Coates -.576
Mitch Williams -.643
Ed Summers -.663

Top 10 pitchers:
Mariano Rivera +1.858
Rollie Fingers +1.207
Jack Morris +1.027
Art Nehf +.979
Allie Reynolds +.859
Curt Schilling +.849
John Smoltz +.844
Herb Pennock +.841
Bob Gibson +.821
Sandy Koufax +.818

That's a lot of names that you'd expect... and Ed Summers.

I'm a WPA whore.

WPA is really only interesting while a game is in progress. Once the game is over it is less than worthless. The sac fly in the ninth is never more important than the grand slam in the first.

WPA is really only interesting while a game is in progress. Once the game is over it is less than worthless. The sac fly in the ninth is never more important than the grand slam in the first.


This.

The sac fly in the ninth is never more important than the grand slam in the first.

As an isolated event, no. But in the broader context of how a player's performance is distributed over the course of a season, or career, it's useful to draw the distinction between performance in high-leverage situations and performance in low-leverage situations. As suggested above, it's useful as a retrospective tool rather than a predictive tool.

Like RBI, WPA is opportunity-sensitive - which is why I prefer WPA/LI to raw WPA.

-- MWE