Mets by WPA

[Edgy MD]

Batters
1) John Olerud 1998: 5.99 (173rd, all time)
2) Darryl Strawberry 1987: 5.91 (188)
3) Mike Piazza 2000: 5.25 (313)
4) Keith Hernandez 1984: 5.23 (316)
5) Darryl Strawberry 1988: 4.98 (375)
6) Bernard Gilkey 1996: 4.91 (390)
7) Mike Piazza 2001: 4.76 (432)
8) Howard Johnson 1989: 4.72 (444)
9) John Olerud 1997: 4.65 (462)
10) Howard Johnson 1987:4.59 (490)

Pitchers
1) Dwight Gooden 1985: 9.93 (1st all time! By a lot!)
2) Tom Seaver 1969: 6.69 (26)
3) Tug McGraw 1972: 6.52 (35)
4) Jerry Koosman 1969: 6.41 (41)
5) Tom Seaver 1973: 6.40 (42)
6) Tom Seaver 1971: 6.03 (59)
7) Tom Seaver 1975: 5.74 (85)
8) Tug McGraw 1969: 5.63 (98)
9) Tom Seaver 1968: 5.08 (150)
10) Tom Seaver 1977: 4.86 (181)

I certainly may be overlooking somebody.

[metirish]

What does WPA mean and why is it significant?

[Benjamin Grimm]

metirish wrote: What does WPA mean and why is it significant?

I was just about to ask that too.

[Edgy MD]

Win Probability Added.

It measures the increased likelihood of your team winning the game, based on your performance, so a three-run homer when down by two in the ninth is worth far more than the same homer in a 14-2 blowout.

It's developing, and the further back we go, the more unavailable (or corrupted) play-by-pay data is, so latter-day players will dominate. They also don't do much of a job measuring defensive contributions, so these may even out as they find a manner to extract fielders' defensive performances from pitching and add it to the the offensive WPAs of batters.

batmags, among others, doesn't like it (and please correct me if I'm wrong) because it measures clutch performance, which he contends is giving players credit for random variation, as clutch performers don't really exist (even if clutch performances do).

I rather like it, because if clutch performance is merely random variation, it should be corrected for in a large data set, and if it isn't, well, here we have a measurement for it.

I further like it because whether or not you hold that some guys are awesome because of their particular ability in the pinch, we have to find a way better than WAR to weight the performances of guys whose entire job is designed to perform in high leverage situations --- relief aces and pinch-hitters. So I appreciate seeing Tug McGraw appearing twice on the above list.

[Ceetar]

whether or not clutch exists, the actual hits themselves are still clutch in that even if it's not an ability, Piazza hitting a three run blast against the Braves to put the Mets up 10-8 in the 8th is still adding a heck of a lot of win percentage.

[Guest Mets � Willets Point]

[Benjamin Grimm]

Is this the measurement that m.e.t.b.o.t. uses in Schaefer voting?

[Edgy MD]

Rusty Staub in 1983 had a WAR of 0.4, but a WPA of 0.9. I would contend that the latter is a more accurate assessment of his contributions.

[Ceetar]

Edgy DC wrote: Rusty Staub in 1983 had a WAR of 0.4, but a WPA of 0.9. I would contend that the latter is a more accurate assessment of his contributions.

I haven't read up on this stat as much as some others, but it seems to me WPA is trying to contextualize WAR. WAR gives you value based on how much things lead to wins. a solo home run is generally worth 1 run and roughly 10 runs is 1 win in WAR, but that solo home run is in the bottom of the ninth of a tie game it certainly seems like it's worth more than .1 WAR and I think that's what WPA is trying to do.

[batmagadanleadoff]

Edgy DC wrote: Win Percentage Added.... batmags, among others, doesn't like it (and please correct me if I'm wrong) because it measures clutch performance, which he contends is giving players credit for random variation, as clutch performers don't really exist (even if clutch performances do).

Since you asked.... I'm with this Ceetar post:

Ceetar wrote: whether or not clutch exists, the actual hits themselves are still clutch in that even if it's not an ability, Piazza hitting a three run blast against the Braves to put the Mets up 10-8 in the 8th is still adding a heck of a lot of win percentage.

WPA is a neat tool to measure the clutchnes of the actual hits. But I don't see much predictive value in the stat. If you look at the list of Mets WPA's you've assembled, it's no coincidence that the best franchise WPA's more or less coincide with the best Mets individual seasons. The best hitters tend to hit the best in the clutch. Because they're very good hitters, not because they're clutch hitters. Who would you rather take your team's last WS at bat with two outs in the 9th, down by four with the bases loaded? Al Weis (1 WS HR; .455 WS BA) or Dave Winfield (0 WS HR's; sub .250 WS BA)? WPA sez to send Weis up there. Do you agree?

Edited May 23, 2012 by Guest

[Frayed Knot]

The advantage of WPA is because of the context, that it measures not just what happened but in what situations.
The drawback is that it's probably not as good a predictor of future success for the very same reasons. I'd imagine WAR WinShares or such similar stuff would be less erratic over a player's career.


oe: cross posted w/Mags

Edited May 23, 2012 by Guest

[Ceetar]

Although it's missing Beltran's 2006.

[Edgy MD]

Sorry, Carlos. Sorry HoJo and Olie.

1) John Olerud 1998: 5.99 (173rd, all time)
2) Darryl Strawberry 1987: 5.91 (188)
3) Carlos Beltran 2006: 5.34 (301)
4) Mike Piazza 2000: 5.25 (313)
5) Keith Hernandez 1984: 5.23 (316)
6) Carlos Beltran 2008: 5.00 (372)
7) Darryl Strawberry 1988: 4.98 (375)
7) Bernard Gilkey 1996: 4.91 (390)
9) Mike Piazza 2001: 4.76 (432)
10) Howard Johnson 1989: 4.72 (444)

Pitchers
1) Dwight Gooden 1985: 9.93 (1st all time! By a lot!)
2) Tom Seaver 1969: 6.69 (26)
3) Tug McGraw 1972: 6.52 (35)
4) Jerry Koosman 1969: 6.41 (41)
5) Tom Seaver 1973: 6.40 (42)
6) Tom Seaver 1971: 6.03 (59)
7) Tom Seaver 1975: 5.74 (85)
8) Tug McGraw 1969: 5.63 (98)
9) Tom Seaver 1968: 5.08 (150)
10) Tom Seaver 1977: 4.86 (181)

[batmagadanleadoff]

Curious: What are Staub I's WPA's? I vaguely recall that he had some WPA seasons that were way over what you'd expect based on his regular season stats?

[m.e.t.b.o.t.]

Benjamin Grimm wrote: Is this the measurement that m.e.t.b.o.t. uses in Schaefer voting?

m.e.t.b.o.t. uses WPA as the primary basis for determining schaefer voting, with crude, rudimentary adjustments made for some defensive trigger events.

WPA measures the outcome of the performances of players in the context of hte game situation. leverage index further evaluates the context of the individual situation by determining, essentially, the impact of the possible outcomes. m.e.t.b.o.t. does not really know how leverage index is calculated, as most online descriptions seem overly simplistic, and merely descriptive, without getting into the analytics.

WPA is certainly not predictive. it certainly is interesting.

[Edgy MD]

batmagadanleadoff wrote: Curious: What are Staub I's WPA's? I vaguely recall that he had some WPA seasons that were way over what you'd expect based on his regular season stats?

I happened to just look at him. Staub through most of his career is an outlier. He leads the league with a 5.7 WPA in 1976, while his WAR of 2.5 doesn't even put him on the map.

[Chad ochoseis]

That's because he's so darn clutchy.

I wonder if it's a function of being a good hitter on a good but weak hitting team. The good pitching keeps the games close, and the one good hitter in the middle of the order is going to be the guy who gets the opportunities for key game-winning hits.

If the team is a generally good hitting team, there's going to be more players to share the WPA. If the pitching and the hitting are weak, the whole team's going to be bad, so there will be more blowouts and the hits the team does get will matter less, WPA-wise.

[Edgy MD]

It's a theory. But I'm not going to jump into it.

If your're the good hitter on a bad hitting team, I would guess that you'd put up a good OPS in clutch situations, getting walked all the time, but wouldn't necessarily get the chance to win the game with a hit, therefore suppressing WPA. Just my way of looking at it.

Let's take late model Barry Bonds --- quintessential good hitter on a bad hitting team. Last ten years of his career.

Season	Off WAR	Off WPA	Dif
1998	7.0	7.0	0.0
1999	4.0	3.1	-0.9
2000	7.0	6.2	-0.8
2001	12.2	11.5	-0.7
2002	11.6	10.5	-1.1
2003	8.4	8.0	-0.4
2004	11.2	12.9	1.7
2005	0.5	0.2	-0.3
2006	3.8	4.8	1.0
2007	4.2	4.3	0.1
Sum			-1.4

Now, it's probably not fair to describe all those Giant teams as generally poor hitting apart from Bonds, but if your theory held, I would expect his WPA to regularly exceed his WAR, but the opposite appears to be true.

[MFS62]

Edgy DC wrote: It measures the increased likelihood of your team winning the game, based on your performance, so a three-run homer when down by two in the ninth is worth far more than the same homer in a 14-2 blowout.

I still recall my Pirate fan friend telling me that Dick Stuart was "the greatest hitter you ever saw in a 12-3 ballgame" when Stuart joined the Mets.

Later

[Edgy MD]

And we all feel that way about somebody, true or not. Hopefully this stat can gauge some truth in there.

[MFS62]

Edgy DC wrote: And we all feel that way about somebody, true or not. Hopefully this stat can gauge some truth in there.

Yes. Interesting stat. I'd never heard of it before this thread.
Later

[Edgy MD]

It's something I've wanted to measure my whole life.

[Benjamin Grimm]

Edgy DC wrote: It's something I've wanted to measure my whole life.

[Edgy MD]

NL Top Ten, Offensive WPA, 2012

1. Votto (CIN) 3.3
2. Wright (NYM) 3.0
3. Gonzalez (COL) 2.4
4. LaHair (CHC) 2.3
5. Beltran (STL) 2.2
6. Prado (ATL) 2.0
7. Nieuwenhuis (NYM) 1.9 (!!)
8. LaRoche (WSN) 1.9
9. Kemp (LAD) 1.9
10. Braun (MIL) 1.8

[smg58]

It's easy to forget that Olerud had some obscenely good seasons. In 1993 with the Jays, he had an OPS+ of 186.

[Edgy MD]

And yet he slumped badly enough that the Jays lost faith and we picked him up in a salary dump.

[Edgy MD]

Steve Henderson's homerun on June 14, 1980 was good for a .919 WPA, taking. Can any single hit in their history top that?

[Ceetar]

what was Piazza's home run valued at 6/30/2000?

[Edgy MD]

0.375. Nothing to sneeze at, but the game was already tied, and the ninth inning was still to come.

[batmagadanleadoff]

Edgy DC wrote: Steve Henderson's homerun on June 14, 1980 was good for a .919 WPA, taking. Can any single hit in their history top that?

You'd practically need a walk-off grand slam to give the Mets a one-run win: the highest WPA value hit possible.