Monday, September 13, 2010

Not all situations are the same

Julio Borbon of the Texas Rangers not only squeezed home
the go-ahead run with this seventh-inning bunt
on Sunday, he reached first base safely.
Aaron Gleeman this morning linked to this tweet from Matt Klasssen of Fan Graphs, which claims that less than a third of bunt attempts increase the bunting team's Win Probability Added.

WPA is a situational stat. It attempts to measure how much a given act (a single, a stolen base, a double play) affects a team's chances of winning the game given the situation in which it occurred. Jim Thome's home run in the 12th inning Saturday ranked a lot higher in WPA than did his home run last Tuesday, when the Twins already had the lead.

And like many of the sophisticated sabermetric stats, WPA is essentially non-auditable, and I regard such stats with varying levels of suspicion.

Gleeman takes Klaassen's tweet to be another nail in the coffin of the sacrifice bunt. I don't — because every situation is different.

Man on first base, nobody out, tied game, ninth inning. WPA views this as the same situation whether the batter is Jim Thome or Jason Repko, whether the pitcher is Bobby Jenks or Matt Thornton, whether the third baseman is Omar Vizquel or Mark Teahen.

No, you don't bunt with Thome; that runner on first is already in scoring position. With Repko, it makes more sense — especially if the erratic Teahen is at third and the chunky Jenks is on the mound. And it makes a bit less sense if Vizquel is playing third and Repko has the platoon advantage against Thornton.

A lot of variables there — and we haven't even discussed weather conditions, home or road, length of grass ...

I've blogged about this specific paragraph from Bill James (in his 2010 Gold Mine book) before, but it remains relevant:

Sabermetricians are often critical of the bunt, arguing that the sac bunt, even when successful, reduces the number of runs the team can expect to score. But this misses a critical point: that the "Denard Span" bunt, where you're really bunting for a hit but you'll take the sacrifice as a by-product of failure, is a very good play. If there's a runner on first and nobody out ... you only have to bunt about .275 to make it a good play — assuming that you'll get the sac bunt even if the effort for a hit doesn't work. A good bunter can bunt much more than .275 —making it a good play.

(James was writing in the context of examining Span's bunting prowess in 2009, when he bunted 27 times for 12 sacrifices and 10 singles. That's a .370 "batting average" when bunting.)

Earl Weaver hated the bunt. Billy Southworth, another Hall of Fame manager — who managed a team Weaver followed as a kid — loved the bunt. Tom Kelly tended to avoid the bunt; Ron Gardenhire deploys it frequently.

This difference of opinion is well and good — and logical. They all managed in different times and places, with different players of various skills.  The illogic comes in envisioning the game as one-strategy-fits-all.


  1. Agreed, Ed. Another piece is who the on deck batter is, whether they are lefty or righty, same with the pitcher... and of course, with 1B open - walk the next batter to keep the DP in order.

  2. Gardy is about average or a little more in bunting. The Twins did lead the league one year, but several years were just below average in sac bunts. TK definitely didn't like the bunt. The Twins were consistently at the bottom of the AL in sac bunts under him despite having completely punchless offenses. I agree about trying to bunt for hits in bunting situations, especially early in games. I wish Punto, Tolbert and Casilla would do that more often.