May 18, 2010

Red Sox/Yankees WPA Leaders Since 1995

ESPN's Mark Simon took a look at the Win Probability Added data for all regular-season Red Sox/Yankees games since 1995 -- the beginning of the wild-card era.

Simon uses one play from June 3, 2007 to illustrate how WPA works. Alex Rodriguez was batting against Jonathan Papelbon (top 9th, tie game, two outs, no one on base). From all of the existing play-by-play data, we know that teams in the Yankees' position at that moment have won 38% of the time. Slappy hit a home run.* In the new situation (top 9th, visiting team up by 1, two outs, no one on base), teams win 78% of the time.

* I was set to post this yesterday afternoon, but held it back when L said her guest post was ready. Bad example in light of last night's debacle!

The Yankees' chances of winning increased by 40%, so Rodriguez gets 40 percentage points (0.40) for hitting the home run and Papelbon receives a debit of 40 percentage points (-0.40) for allowing it. When you do that for every play in every game, you get a total of how much a player increased or decreased his team's chance of winning.

Note: WPA measures only hitting and pitching, not base running or fielding (or the calmness of one's eyes). And perhaps because of that oversight, CI is merely 13th on the Yankees' list of most productive hitters against the Red Sox over the last 14 years -- with less than one-half of one win to his credit -- falling in comfortably behind Enrique Wilson.

Worst hitter and pitcher for each side: Shane Spencer and Jose Contreras; Jason Varitek and (by a huge margin) Derek Lowe. (Second-worst Boston hitter and pitcher? J.D. Drew and tonight's starter Josh Beckett.)


Joe Gravellese said...

WPA is a weird stat. I think it's a lot of fun, and I like reading up on it and checking out the graphs... as long a you take it with a grain of salt.

mattymatty said...

This is why you're a great blogger, Allan. If this were my blog you wouldn't read a damn thing about last night. Not because I'd be trying to hide it but because I just don't want to think about it - its too depressing.

Thanks for all the work you do here. Its appreciated.

allan said...


It does exactly what it sets out to do -- tell you how much (or how little) each player contributed to a win or loss. Not a wonder stat to replace everything else, but pretty ideal.

And it rightly weighs a fourth-inning single in a tie game as *way* more important than a grand slam in the 9th inning of a blowout.

Thanks, mm.