In our last article, we saw that the new flat-seamed ball has led to an increase in scoring in college baseball. If you're used to Major League Baseball, you might be relieved, since more scoring means runs are easier to come by, which in turns means teams should start moving away from small ball tactics such as stealing too often and bunting. Especially bunting. Man, do sabermetric people hate bunting.
But that's professional baseball. Just because Manny Machado knows what to do when the batter lays one down doesn't mean bunting is stupid across all space and time. If you're at a lower level, where the fielders aren't as skilled, a bunt could be a good call: there's a better chance the infielders throw the ball away, or someone forgets to cover a base, and so you end up advancing the runners without actually giving up an out.
The simplistic way to determine when bunting makes sense is to look at a run expectancy table: Since having a runner on first with no outs is worth more runs (on average) than a runner on second with one out, most people will conclude that bunting in that situation doesn't make sense. But that's not an accurate description of the decision managers face. A more reasonable approach would be to compare the average run expectancy after bunting with the average run expectancy after swinging away for each base/out state. This will take into account both the myriad outcomes that can follow this decision (e.g., home runs, dropped third strikes, bunting into triple plays) and their relative frequencies.
But even this method is a simplification. We're still assuming an average batter facing an average pitcher; having a really good pitcher face a really weak hitter, for example, makes a positive outcome while swinging away less likely. And, as we will explore in a later article, different conferences have different talent levels and run environments.
For now, though, let's look at the average outcomes across all conference games. This won't account for the disparity in the batter and pitcher talent levels, but by limiting ourselves to conference matchups, we'll at least eliminate the influence of those one-sided early season games. Any play labeled as a bunt in the play-by-play database was included. As a result, we're obviously also inadvertently counting batters who were bunting for a hit. Some of these — such as those that came with the bases empty or with two outs — can obviously be excluded, but everything else is impossible to differentiate. This method might also tend to undercount strikeouts on foul two-strike bunt attempts, if the scorer keyed in the third strike as a swinging strike instead of a foul bunt.
This table shows the difference between the run expectancy when bunting with the run expectancy when swinging away in the base/out situations where teams bunted the most. To judge the efficacy of a sacrifice bunt, we've calculated two statistics. The first, Delta Run Exp, is the average improvement in run expectancy by bunting as opposed to swinging away. (In situations where swinging away produces more runs on average, this number will be negative.) The other, Delta P(score), tracks the change in the probability that a team will score at least one run in a given inning. This column was included because, in close and late situations, a team might be willing to play for a single run.
|Bases||Outs||Bunt PA||% Bunts||Delta Run Exp||Delta P(score)|
Just like in MLB, bunting in the most common situations produces next to no change in run expectancy, or (in the case of a runner on first with no outs) actually produces fewer runs on average. But bunts in less common situations, where maybe the defense is back on their heels, do produce higher run expectancies on average. And in almost all cases, the probability of scoring at least one run increases. For example, teams that bunted with a runner on second and no outs scored at least one run in 68 percent of innings, whereas teams that swung away in that situation scored 62 percent of the time.
How does this compare to MLB? This table covers the same base/out situations shown above in all games from 2010 through 2014. In contrast to the college baseball numbers, MLB teams that bunt score fewer runs on average than those that swing away, and actually score in a lower percentage of innings in many cases.
|Bases||Outs||Bunt PA||% Bunts||Delta Run Exp||Delta P(score)|
It's worth mentioning, especially as we consider the MLB numbers, that the group of hitters who get to swing away will be very different than the group of hitters that are told to bunt. Not only that, but the stronger hitters that are swinging away are more likely to be surrounded by other strong hitters, who in turn will be more likely to produce more runs than their counterparts lower in the order.
Then again, the same can be said for college teams: Kris Bryant probably didn't get the bunt signal too often at the University of San Diego. But there we do see an improvement in the likelihood of a single run scoring, and (in some situations) the average number of runs scored. For further evidence of this, let's look at the percentage of balls in play that resulted in the batter reaching safely (excluding fielder's choices, since those don't advance the runners). It's sort of like BABIP, but since it includes reaching on errors, I'll call it rBIP: reach percentage on balls in play.
|Bases||Outs||D1 rBIP||MLB rBIP|
As you probably expected, the NCAA rBIP is higher than the MLB rBIP in every base/out state. This suggests that bunting is a more reasonable option for college coaches than for MLB managers, and the improvement in rBIP suggests this is due to the improved defensive skill of major-league defenders.
It might be tough, but analytically-minded fans watching college baseball are going to have to grit their teeth and learn to appreciate the art of the sacrifice.
. . .
MLB statistics courtesy of Retrosheet. College statistics courtesy of NCAA, and are freely available in MySQL-compatible format through Bryan's GitHub page. Special thanks to Christopher D. Long and Meredith Wills for their web scraping code, and Jeff Wiser for his feedback.
Bryan Cole is a featured writer for Beyond the Box Score. He really wanted to title this article "How I Learned to Stop Worrying and Love the Bunt" but it exceeded the character limit.You can follow him on Twitter at @Doctor_Bryan.