It is officially February when baseball news is reduced to vague rumors about teams from which Roy Oswalt and Edwin Jackson may or may not be considering one-year offers. Well, that and that Mark Teixeira saying he might bunt to beat the shift this season. Hoo boy.
There is something of interest in the Teixeira report, though. Sure, we do not know whether he is actually going to do it or not. Remember, this is the time of year when players say things like “I’m going to steal 20 bags this year” even if they have never stolen more than 10 in any season. Still, it is not a crazy idea. While sabermetric writing on the internet went through a phase of arguing that bunts are counterproductive to scoring and winning, research has progressed to show that bunts are not as bad as all that. In certain situations, they can be a good idea in terms of getting the win in a close game or simply “keeping the fielders honest” (also known as “game theory,” a term I am pretty sure Bruce Bochy uses frequently).
But what about power hitter like Teixeira? Isn’t bunting always a bad idea for them? To answer this properly would require a great deal of complex thinking and programming. For now, let’s take a simple approach by looking at some data from 2011 to see whether Teixeira is simply blowing smoke or making sense.
One approach would be to look at different situations in which bunts might be advisable or not advisable for particular hitters, and that is obviously the way that the hitter and his team should approach it. However, I want to take a more general, introductory approach to looking at the productivity of bunts by sluggers in general. It is important to keep in mind that many the background assumptions regarding good bunting situations, as I will briefly mention again at the end.
I have posted seasonal reviews of bunting in the past, but those were primarily focused on the managerial-strategic value of particular bunts, so I used Win Probability Added in those cases. My intent here is different. I want to see how a particular subclass of hitters did when bunting in 2011. I will use RE24, or the average change in run expectancy on bunts. Without getting into a long-winded explanation of RE24, I will simply recommend this entry on linear weights) from the Sabermetric Library.
In short, based on the expected runs scored by the end of the inning for each of the 24 base/out states, RE24 measures the change in run expectancy for each plate appearance — that is, the base/outs state’s run expectancy at the beginning of the plate appearance, after the plate appearance, and including any runs scored. This is also the empirical basis for commonly used metrics such as offensive linear weights (the basis for wOBA). The average RE24 of an event is its “linear weight,” e.g., the average RE24 of a home run is usually around 1.4 runs in modern baseball. Like WPA, RE24 helps us getting a better perspective on particular bunts than by simply repeating “making outs is bad.”
Let’s check the database (with my iffy SQL skills in mind) to see what some different kinds of bunts were worth in 2011. (Ideally, yes, I would use multiple years of data and make other adjustments, but I am trying to keep this simple for the purposes of this post.)
The average run value of a bunt in 2011 was -.0571 runs, or less than one-tenth of a run worse than the average plate appearance. That includes all bunts, not just sacrifices, and also includes pitchers bunting. (The average value of a sacrifice bunt in 2011 was -0.1153.)
Let’s narrow things down to something closer to Teixeira’s actual population. For non-pitchers, the average bunt in 2011 was worth -.0083 runs. (The average sacrifice bunt by non-pitchers in 2011 was worth -.0890 runs.) You may have noticed (without me doing a separate query) that the average value of a pitcher bunt is lower than that of non-pitcher bunts. Does this mean that pitchers are bunted too often or in the wrong situations in 2011? Well, maybe, but not necessarily. While pitchers probably are bunting in situations that are less bunt-friendly compared to the average, keep in mind that the vast majority pitchers are far, far, worse than non-pitchers when hitting, so bunting is acceptable and even advisable for pitchers in a wider variety of situations than it is for non-pitchers.
That digression aside, what about power hitters bunting in 2011?
[In case I did not make is clear before, I am not doing this the "best" way. Ideally, I would have the probabilities (estimated from true talent in 2011) for each individual hitter's potential events (strikeout, ground out, walk, single, home run, etc.) for each batter to compare those to the change that came from bunting in those situations. More simply, one might at least want a sub-Marcel true-talent wOBA estimation to compare the average expected change in run expectancy for each plate appearance compared to a bunt. Moreover, multiple years of data would be better, too. All of that is simply to acknowledge that what I am doing -- using observed performance from one season of data -- is obviously limited. However, I think it is interesting at the very least as a spur to further work in this area.]
What I did was simply to look at the average run-value of bunts for two groups of non-pitchers in 2011: those who had an isolated power of .200 or greater, and everyone else. I limited the query to hitters with at least 300 plate appearances in 2011. (Note that the earlier queries did not set plate appearance minimums, so these values are not going to add up to those given earlier).
The sluggers’ (those with an observed ISO of .200 or greater in 2011) average bunt was worth +.0376 runs. The sample was only 102 bunts, as one might expect.
(As an aside, I realize that .200 is an arbitrary figure. A .180 ISO is still good power and would have made the sample larger, but on the other hand .200 and better reflects Teixeira’s population of power hitters.)
How about the non-sluggers, those non-pitchers with 300 or more plate appearances in 2011 and an ISO under .200? They bunted 1476 times according to my query, with an average run value of -.0023.
Does this mean that the sluggers are better bunters? Should they be bunting more often? To the first question I would say that I do not know, but I seriously doubt it. As for whether they should be bunting more often, that would require a more work. This is where the previously-mentioned “background assumptions” come into play. Sample size aside, and without looking at individual plays, I think it is fair to assume that the typical “slugger bunt” comes as more of a surprise than the non-slugger bunts, and that teams were playing back or had a shift on for the power-hitter.
That is not to say that teams should not continue to shift or play back — after all, it was just 102 bunts, and while it might be frustrating to have a power hitter reach on a surprise bunt once or twice a season, it beats playing in and/or without a shift and giving up several more doubles.
Like many other fans, in the past I have wondered why more power hitters do not try to bunt to the opposite field more often to beat the shift. And, of course, like so many others, I needed to be reminded that if it was that easy, they probably would. I would imagine that Mark Teixeira is not the only one of his homer-heavy peers who has not bunted since high school. Moreover, they probably are not going to be able to do it often enough to get teams to play in or give up the shift.
However, while one might chuckle at the idea of Teixeira being a “$180 million bunter,” our little foray into recent data shows that whether or not having a slugger lay down a bunt now and then is good from a game-theoretical perspective, it can actually help teams score runs.