The Wins and Losses of the Sacrifice Bunt

By Robert Pike
7 months ago


“They can take the analytics on that and shove it up their [bleep][bleep].” – Ron Washington on bunting

Unfortunately, Gil LeBreton of the Fort Worth Star-Telegram censored Rangers Manager Ron Washington, but I’m going to assume the bleeped out phrases were “internet” and “blog.”

If there is a single event in baseball that most separates the sabermetric community and the old-school community, it might be the sacrifice bunt. Baseball researchers have developed “run expectancy matrices,” like these, illustrating that “sacrificing” an out to advance a runner often-times decreases a team’s expected runs scored for that particular inning. Baseball Prospectus has a fantastic tool for this as well.

image 1

 This chart, tweeted by Scott Lucas (@scottrlucas), demonstrates the historical probability of scoring each number of runs in the current (4.3 runs/game) environment by advancing a runner from second to third while sacrificing the first out of an inning.

Essentially, the findings of baseball researchers have concluded that while the sacrifice bunt does increase the team’s odds to score a single run in certain scenarios, it does not maximize a team’s run-scoring probability.

So, while many can agree that a sacrifice bunt might be a sound strategy in the late innings of a tight game, it seems to drive the sabermetric community insane whenever a manager attempts to deploy this strategy in the early innings.

FanGraphs’ Jason Collette asks an interesting question: how much does bunting early in games hurt teams like the Rangers over the course of the season? The answer actually surprised me.

These 29 sacrifice bunts called by Ron Washington in the first four frames occurred in 27 games.

The Rangers were 20-7 in those games.

Three of those 29 sacrifice bunts were laid down by pitchers, leaving 26 by non-pitchers.

image 2

 Non-pitcher sacrifice bunt attempts by Rangers’ players (first four innings of a game in 2013)

In games in which position players attempted a sacrifice bunt in the first 4 innings of a game, the Rangers were 19-5. They averaged 7.08 runs in each of these games. Per inning, the Rangers averaged about .5 runs per inning for the entire year. Within the 26 innings included in this sample, they scored 39 runs, or 1.5 runs per inning!

Of those five losses, two of them came by a margin of one run, one came by a margin of two runs, and two of them came by a margin of four runs. Only two of the games won were decided by two runs, with the remainder of the games being decided by three or more.

Worth noting is that the run expectancy matrix, naturally, is a league-wide average. This means that it assumes that the player laying down the bunt is the average run-creator. Of the players asked to lay down a sac bunt for Ron Washington in 2013, only Craig Gentry and Ian Kinsler (who totaled four early-inning sac bunts between them) finished the season with an OPS above the league average.

Irrespective to inning, and including pitchers, the Rangers deployed 65 sac bunt attempts in 2013, and scored 68 runs in those 65 innings. Of those 65 sac bunts, 54 of them were with a hitter at the plate with a below-league-average OPS.

It’s a very small sample, especially given the slight changes in probabilities, but it would seem to indicate that Ron Washington could be suffering some confirmation bias – it does seem to work for him (in fact, according to Baseball Reference’s play index during the Ron Washington era the Rangers have laid down 467 sacrifice bunts and scored 521 runs in those innings).

To answer Jason’s question, I’m surprised to say – they would likely be in the exact same spot.

In fact, doesn’t this show that Ron Washington’s club had a greater run expectancy after deploying a sac bunt? The math tells us the sac bunt surrenders fractions of a run almost every time one is put down, but during his relatively brief time at the helm Ron Washington has beat the odds in a small sample, and it appears he has no intent (or reason?) to steer his club in a different direction.

How does the rest of the AL fare when a non-pitcher attempts a sacrifice bunt during the first four frames? One hundred ninety non-pitcher bunts were laid down by AL teams in the first four innings of 178 different games in 2013. AL teams were a surprising 109-69 in games that they employed a sacrifice bunt under those circumstances. Those teams scored an average of 5.23 runs per game in those games, nearly a full run above the AL average runs per game of 4.33 in 2013!

While this is still a small sample, and I still believe that the run expectancy matrix would tell us that this type of behavior will not continue to see this amount of success, is it also possible that the outrage over the sacrifice bunt might be over-sold?

Robert Pike is a Contributor at Paranoid Fan. Follow him on Twitter at @Bob_Pike.