A common occurrence this year has been TLR using position players to sac bunt guys to third after a leadoff double. Generally speaking this move is killed by the sabermetric community and at least occasionally lauded by the old-school small ball community. This post is going to be less about who’s right and more about under what set of assumptions/circumstances would each particular side be correct.
First let’s talk about the specific assumptions that need to be made.
- Run environment – high scoring or low scoring. Generally something like 4.4 runs/team is average.
- Probability of a successful sacrifice. MLB average is ~71%.
- Probability of a hit/error on the attempt
Those are the big three that we will be dealing with throughout the post. We’ll parametrize all of those assumptions and see where the “tipping points are”. The first chart assumes a 75% success rate and parametrizes the other factors
The “Hit Away” line is the run expectancy (RE) entering the at-bat (i.e. runner on 2nd 0 outs) and each other line is the RE for the given Hit% (actually it is the hit+error%). Basically with anything less that a 25% hit percentage the best approach is to hit away; which I would think covers most batters. With a 25% hit percentage bunting would be the best course of action only in the lowest of run environments (where your team only expects to score ~2 runs. Clearly this is almost never the case in the modern game (maybe Pedro in his prime?). In the modern game (~4.4 runs) the break even point is about a 30% hit percentage.
What if we assume a better bunter? I originally started this research after Tony bunted with Jay, and noticed that he had been successful in 86% of his sac bunt opportunities. Clearly the sample size is so small we cannot conclude that this is his true talent; however let’s run the number assuming it was. The following chart summarizes
Under these assumptions anything less than a 20% hit percentage doesn’t really add value. At a 20% hit percentage bunting is a break even strategy in a run environment between 2.5 and 3 runs per game. In the modern game (~4.4 runs per game) the break even point is a 25% hit rate.
Draw from these results what you will. I’ve called out the break even points, so you can decide which set of assumptions best fits the particular scenario. I will say that these were done under the lens of it being early in the game and thus used RE instead of win expectancy (WE). For later in the game a WE analysis would likely be better as raising your chance to score one run may outweigh scoring multiple runs in certain situations.
RE by run environment info found here via Tango.