THE BOOK--Playing The Percentages In Baseball

AND export as Excel or CSV.  It’s truly been a pleasure working for David.

Yeah, David is superfast on the response, and very accomodating.

Fangraphs now has WPA since 1974, excluding 1999.

Here’s Tim Raines:
http://www.fangraphs.com/statss.aspx?playerid=1406&position=OF

His WPA/LI is +42 wins
REW = +55 wins
WPA = +51 wins

And Mike Schmidt:
http://www.fangraphs.com/statss.aspx?playerid=1011586&position=3B
WPA/LI = +64 wins
REW = +65 wins
WPA = +54 wins

So, what does this mean?  If you look only at REW, that’s Linear Weights by the 24 base/out states.  To the extent that you think someone can tailor his approach by base/out, REW is what you want.  Schmidt is 10 wins ahead of Raines.

WPA/LI is situational wins, which also include inning/score.  To the extent that you think someone can tailor his approach to the game situation (and making sure each PA counts the same, i.e., each PA is equally weighted as 1), then this is what you want.  Schmidt here has a 20 win advantage on Raines.

WPA is change in win expectancy.  And, the closer the game, the more each PA is weighted.  So, in a close game, a PA could get 5 or 10 times the impact of a random PA.  This one is the one that exactly matches a team’s W/L record, and assigns all the benefit (and disadvantages) to the players for that play (similar to being lucky enough to buying an IPO as an insider… you didn’t do anything special, other than be smart enough to lay your money out).  If you believe that Mariano Rivera should get a big boost for being used in high-pressure situations, then this is the one you want.  Schmidt was a horrible clutch hitter, while Raines was fantastic, enough that Schmidt only has a 3 win advantage here.

If you just want a straight Linear Weights, baseball-reference.com has that for you as BtnWins, and it comes straight from Palmer I believe.

Choose whichever floats your boat.