Tuesday, April 19, 2011

Batting Average on Balls in Play

There's this mythology that batting average on balls in play is a matter of pure luck. The pitcher, so the thinking goes, determines whether an at-bat results in a walk, a strikeout, a home run, or a ball put in play; the first three have deterministic outcomes, and in the last case it's just a roll of the dice. Moreover, according to the hardest-core statistical minimalists, there aren't fundamental differences among pitchers in their BAbip levels. There aren't low-BAbip pitchers, there aren't high-BAbip pitchers, because BAbip is not a pitching skill. Your skills are striking people out and preventing home runs or walks. That's it.

On his career, Nolan Ryan's BAbip was .266, distinctly below the league average over that period. Never mind that he's enough below league average over a long enough career that I think it's almost certainly a statistically significant difference. The standard deviation of his BAbip across the couple dozen season he pitched was .021, or right around 8% of the number itself. Now consider that his career strikeouts per plate appearances number was just a little less than 25% (!) and the standard deviation of his K/PA number across various seasons was .036, which is 14.5% of the number itself. In other words, Nolan Ryan's propensity to strike people out varied a lot more than his propensity for giving up hits on balls in play. To me that suggests that BAbip isn't as mysterious and uncontrollable as people sometimes maintain. To me it says that there was something about Nolan Ryan's stuff that made it hard to put in play well. In fact, when I originally encountered the idea that BAbip is not a pitcher's skill it was in an article that seemed to rather explicitly negate itself, by noting that knuckleballers, of course, have naturally low BAbip numbers.

Okay. If a knuckleball can induce poor contact, why is it theoretically impossible that a normal pitcher could know how to induce poor contact? Likewise, given that home runs are just balls put in play but a little further than usual, why shouldn't a pitcher who can keep home runs from happening theoretically be able to prevent other kinds of hard contact? I think what we have here is a case of square peg-ism. A lot of what goes into BAbip is luck, certainly. And it's damned difficult to measure whether a pitcher really has a low BAbip as an inherent quality of their own or whether it's just that they're getting lucky. You could watch every play, or have some sort of batted-ball tracker tell you when people give up hard line drives or little dribbling ground balls. Then you could, perhaps, quantify whether a pitcher does genuinely give up hard contact or whether it's just a lot of ground balls finding holes, etc. But they don't yet have a system to do that; similarly they don't have a good way to measure a fielder's range, other than raw "plays made" numbers. So what do you do when there's an element of skill you can't measure easily? Well, you assert that it doesn't exist. Can't measure it, it must not be real.

No thank you.

No comments:

Post a Comment