THE BOOK--Playing The Percentages In Baseball

Don’t forget that 0.8 of Favre’s 2.4 WPA came on one play (game-winner vs San Francisco); he’d be 0.13 WPA per game without that.

The only scenario in baseball I can think of with a possible 80% swing in WE in bottom of the 9th, down 1 with a man on first and two outs. WE is .093, and a HR is worth +.87 wins.

And the NFL only plays 16 games!  Having some fun with pro-rating, .20 WPA/game is 3.2 WPA/NFL season, but **32 WPA** over 162 games.

It seems to me that WPA works well at a team level for football, but that it doesn’t do anything to break down the individual contributions of players.  The author talked about the major flaws in assessing both defenders and O-linemen, which seem like pretty important issues with applying WPA to a player evaluation like this.  It also sounds like every time a player is mentioned in a play, he is credited or debited the full WPA of that play, even if that means giving out 2 or 3 times the actual WPA of the play to players.  This especially seems like a problem for QBs/receivers, in that you’re always going to be double-counting the value of the play and giving everyone too much credit as a result (unless I’m misunderstanding how WPA is applied to players).

If Favre got .8 WPA on one play, and the receiver also got .8 WPA on that play, then that is 1.6 wins given out on a single play without even considering the contributions of the blockers or other receivers who bought time or occupied other defenders.  When WPA is applied to players in baseball, while individual contributions aren’t always disentangled perfectly, you can at least total up the individual contributions into a total for the team.  If I’m reading this right, the sum of individual totals won’t make any sense on a team level here, which kind of defeats the purpose of using WPA to begin with.