It might only look that way. See Clutch, WPA/LI, and the Home Run Bias by Kincaid of 3-D Baseball and HR Bias in Clutch scores, based on Situational Wins (WPA/LI) by Tangotiger.
I posted a comment on Kinkaid's post:
"Very interesting. Thanks for mentioning my research. What I posted at tango's blog I went into more detail on mine. See
Don't Let Your Little Leaguers Grow Up To Be Right-Handed Power Hitters Who Strike Out Alot Because They Might Choke In the Clutch.
I also did something several years ago called "Do Power Hitters Choke in the Clutch?" it is at
Do Power Hitters Choke in the Clutch?
A study called “Clutch Hitting: Fact or Fiction?” By Andrew Dolphin suggests that they might. It is at
Clutch Hitting: Fact or Fiction
I also did something similar to what you did here by just comparing WPA to linear weights in my presentation on clutch in 2002 at the Boston SABR convention. I had Bonds doing better in the clutch than his linear weights stats would predict.
"A COMPOSITE CLUTCH HITTING STAT
A STAT THAT TAKES INTO ACCOUNT EVERY PLATE APPEARANCE, NOT JUST CERTAIN SITUATIONS. ONE EXAMPLE IS “GAME STATE VICTORIES” BY RHOIDS SPORTS ANALYSIS
"Game State Victories (GSV) is the number of wins a player contributes to his team by measuring the change in the probability of his team winning the game as a result of his At-Bat. It gives greater weight to hits made when the score is close late in the game than those made when the score difference is large or early in the game."
GSV may be similar to Bennett’s PGP (player game percentage) and the Mills brothers’ Player Win Average.
Results from linear regression
Dependent variable: GSV FOR 2001 (231 players with 300 or more at-bats)
Estimated coefficient values for independent variables (all were statistically significant)
1B = .066
2B = .079
3B = .120
HR = .122
BB = .055
Outs = -.021
R-SQUARED = .819 Intercept = -.79
The .819 means that 81.9% of the variation between hitters in GSV is explained by the 6 independent, nonclutch variables. So 81.9% of the weighted clutch performance is explained by nonclutch performance. At most, 18.1% could be explained by clutch ability.
Pred = the GSVpredicted by the regression equation:
-0.79+ 0.066*1B + 0.079*2B + 0.12*3B + 0.122*HR + 0.055*BB – 0.021*Outs
Outs are At-bats – Hits. Diff = the difference between actual GSV and predicted GSV