I have alluded before to the theory that pitchers have very little control over their BABIP statistic (Batting Average on Balls In Play). In semi-plain English, the theory suggests that once a ball is hit into the field of play, the traits of the hitter – not the pitcher – are what most influence the ball’s likelihood to become a hit or an out. The power of this theory, if it is true, is that pitchers’ BABIP statistics can be expected to level out toward the league average, about 0.300. For example, a pitcher that gives up a 0.350 average on balls in play through the first half of a season is likely to see his BABIP stat drop, and with it his OPS (on-base % plus slugging %).
In a recent study, I looked at the predictive ability of pitchers’ past BABIP statistics on their next-season BABIP statistics, and correlations between BABIP and other stats.
The first part of the study returned some expected results. The following chart shows both the career average BABIP stats for pitchers going into the year in question, and the resulting average BABIP from those pitchers.
Career Babip Next Season
<0.280 ————————-> 0.287
0.280 – 0.299 —————->0.298
0.300 – 0.319 —————->0.299
And then the predictive ability of just the most recent season on next year’s BABIP…
<0.280 ——————> 0.289
0.280 – 0.299 ———>0.299
0.300 – 0.319 ———>0.300
What this tells me is that, for the middle 65% of all pitchers, BABIP is a relatively uncontrollable statistic and tends to level out to the league average. However, the upper 20% and the lower 15% are either “good” or “bad” enough to actually have some slight impact over their own BABIPs. I want to stress that the influence that these two tiers of pitchers have is pretty small. In the study, season BABIPs ranged from 0.203 to 0.380, and the standard deviation was 25 points, yet the “control” that the extreme pitchers had was less than 10 BABIP points. Also, if a player pitched in front of a great defense in a good pitchers’ park, his BABIP is likely to have been on the low side by no skill of his own, yet our data cannot differentiate between these things. Therefore luck seems to have played a larger roll in the variance of BABIP than the “skill” (or lack thereof) demonstrated by the extreme 35%.
So what exactly is it that allowed certain pitchers to have that little bit of control over BABIP? Correlations from over 500 pitchers in 2008 show that strikeout rates and team UZR ratings (a defensive barometer) were the most correlated. Strikeouts are obviously something that pitchers can control, and might help to explain the small influence high strikeout guys like Johann Santana have over BABIP. The correlation to team UZR ratings logically demonstrates that better defenses make more plays and reduce a pitcher’s BABIP. Since correlations were weak for all stats tested, it appears as though we can say that pitchers in general have very little control over BABIP, and its expected regression toward 0.300 can help to predict a pitcher’s performance the rest of the season.