The following is an attempt to explain a puzzling fact about baseball. Mets fans may recall Gary Cohen glibly dismissing the emphasis that SABRE analysis places on strikeouts in assessing pitchers, on the grounds that obviously "a good pitcher will be able to produce weaker contact and thus maintain a low BABIP through skill". And common sense says he should be right.
Except that he is dead wrong.I don't need to tell you he is wrong. A very cursory glance at the BABIPs of various starting pitchers is sufficient to discern that BABIP is overwhelmingly a measure of luck.
I could go on, but it 's easy for anyone to prove to his satisfaction that great pitchers do not have the ability to produce poor contact so as to lower their BABIP, counterintuitive as it may seem. The only aspect of BABIP under a pitcher's control is their ability to produce fly balls, which are converted into hits less frequently than ground balls.
But why is this the case? I suspect the answer lies in three areas: the very real difference between Batted Balls and Batted Balls in play, the influence of the hitter in batted balls, and statistical noise.
First, the category of batted balls in play excludes the weakest class of batted balls (fouls, on average) and the strongest class (home runs). In a sense, this produces an effect similar to a statistic that purports to measure whether or not I can punch as hard as Mike Tyson that excludes Tyson's hardest punches and my weakest. You get a measurement that would probably underestimate the difference between our respective punching abilities.
Consider the following at bat: a strong pitcher--let's call Him Pedro--over the course of five pitches, produces two weak foul balls, a swinging strike, and two balls. On the sixth pitch, he throws a subpar--for Pedro--changeup that results in a moderately hit ground ball.
Now consider this appearance, a weak pitcher--we'll call him Mike--over the course of three pitches, produces two balls and a called strike before throwing his best slider that yields a similarly hit ground ball.
Over the course of a season, these two very different sequences of events repeated over time produce identical results in terms of BABIP. The differences between the two pitchers, on the other hand, tend not to show up in that statistic, but they emerge in other statistics that ostensibly have little to do with balls in play.
For example, Mike's inability to produce swinging strikes, combined with his mediocre fastball, leads to hitters' counts, walks, low K totals, and gopher balls--and possibly extra-bass hits. (There seems to be no mainstream record of how many XBHs a pitcher gives up.) It does not necessarily effect his BABIP; it just results in a greater proportion of his pitches falling into a threshold that can result in a hit.
Meanwhile, Pedro's ability to generate foul balls results in two strike counts and is statistically visible in strikeout totals; but the foul balls fall below the contact threshold registered by BABIP. Similarly, they fall above the threshold of contact required for a strikeout (essentially no contact). The problem is that every plate appearance will ultimately be judged on the final event resulting from it, and there is an entire range of contact that does not yield such a final result. The plate appearance will ultimately be decided on whether or not the pitcher can find his ideal range of contact (the extreme miss) before falling into that BABIP register.
The second aspect to consider in deciphering this curiosity about BABIP is the hitters.
BABIP is just as notably under the control of hitters as it is out of the control of pitchers. Batting accuracy is just as difficult as power, resulting in hitters characteristic Line Drive rates, just as they have characteristic HR/FB ratios. If a pitcher is at all competitive in the major leagues, hitters will remain true to their line drive rates, just as they remain true to their home run rates. The main difference will be how many chances the batter gets to hit those line drives--i.e. it will show in the amount of balls put into play, not in the ratio of the results therefrom.
The crux of the matter here is that it's hard for a pitcher to get major league hitters out, and it is likewise hard for hitters to hit major league pitching. Even a replacement-caliber fastball is hard enough to hit that the results will vary tremendously depending on the skill of the hitters, and it is this differential in skill between hitters that is the main component of a pitcher's BABIP (other than luck). However, if a pitcher does not have an "out-pitch", even a replacement-caliber hitter should be able to stay in the at-bat long enough to generate some contact in the BABIP threshold.
This theory seems consistent with what we really observe in the baseball world. When we see a successful pitcher, we usually see a pitcher who has a good enough fastball to generate foul balls (and even some swinging strikes) to get into a pitchers' count set a batter up for an out-pitch. If a pitcher lacks an out pitch, as we saw with Mike Pelfrey and John Maine, even the nastiest fastballs will simply result in an endless run of foul balls that will keep going until a batter puts one in play. Major league pitchers (in general) are good enough so that even the best hitters cannot generate strong contact consistently enough to reliably produce base-hits. But major league hitters are good enough that a strong out-pitch is required to get them to swing and miss so as to finish them off.
If there were a pitcher in the majors bad enough to consistently yield well-struck balls in play, such as would result in a naturally high BABIP, this pitcher would probably get very few strikes, swinging or looking, very few foul balls, and a very low proportion of pitchers' counts. In other words, it would be obvious that he was not a major league pitcher.
This theory can be tested by observing the amount of foul balls issued by pitchers of various degrees of skill and by including foul balls in batted ball statistics. For all I know, the inquiring baseball minds may have already figured out why BABIP is constant and I might be beating a long-since fossilized horse. I find it interesting, however, that this analysis is consistent with themes that have popped up in the sabre field many times. For instance, when you register an event as a non-event (whether it be with walks in calculating batting average or foul balls in pitching statistics) you are liable to draw misleading conclusions; and when you judge by results and ignore process (whether it be focusing on outs and ignoring how these outs are produced, or recording the result of a plate appearance while ignoring the battle that led to that result) you are liable to get an incomplete picture of what's going on.