A Statistical (Not Sabermetric) Look At Jeremy Hellickson’s 2013 Struggles
In 2011 and 2012, Jeremy Hellickson became renowned throughout the sabermetric crowd of baseball fans for his low batting average on balls in play. Batting average on balls in play (BAbip) is batting average against but taking out strikeouts and home runs, essentially measuring what percentage of balls put into play against a pitcher fell for hits. The league average has traditionally been around .300. In 2011, Hellickson’s BAbip was just .224, and in 2012 it was still outstanding at .264. Suddenly this season, though, it stands at .301, a bit above the MLB average of .296, and Hellickson’s “loss of luck” (as some sabermetricians would term it) has been a crucial piece in Hellickson’s disastrous season. If you look at Hellickson’s BAbip and peripherals statistics the last few seasons, the stats paint Hellickson as a decent pitcher getting very fortunate on the balls hit into play against him the last two seasons, leading to his ERAs that ranked in the top 10 in the American League in both 2011 and 2012, before finally coming back down to earth this year. Is that really what’s going on?
Hellickson supporters have always made the counterargument that Hellickson has forced such low BAbips because he does an excellent job commanding his pitches and forcing weak contact. By that logic, Hellickson’s 2013 struggles are more from worse command than any sort of statistical regression. But that’s a discussion for another time. Right now let’s look at Hellickson’s batting average on balls in play trends through the lens of statistics.
This season in Major League Baseball, there are 93 pitchers who are currently eligible for the ERA title with at least one inning pitched per team game. Among those pitchers, they have a .292 average BAbip with a standard deviation of .0237. To make sense of that .0237 number for the non-statisticians in the room, it tells us that approximately 68% of the pitchers should be within the range of .2683 and .3157 (.292+.0237 and .292-.0237) and about 95% should be within the range of .2446 and .3394 (.292+2(.0237) and .292-2(.0237)). (For you skeptics out there, if you look at the list you can see that 62 out of 93 (67%) are in that first range and 89 out of 93 (95.7%) are in the second.) With the .292 mean and the .0237 standard deviation, we can get down to business trying to make sense of what Hellickson has done the last three years.
Hellickson’s BAbips have been .224, .264, and .301 the last three seasons, giving him a BAbip of .259 overall. Hellickson has managed some extreme BAbip values, but the craziest part about his BAbips have been the range between .224 and .301, a difference of .077 or basically the difference between the lowest in baseball and the average. If we simulated a normal distribution with a mean of .292 and a standard deviation of .0237 a total of 5000 times, the proportion of pitchers that would have a range of .075 or higher between their highest and lowest values is arond 6.5%. When we limit the sample to a difference of .075 or higher and a mean BAbip less than .260, it goes all the way down to just .01%. The probability of a pitcher doing what Hellickson has done over three seasons if if their true mean BAbip should have been .292 is just .0079 or 127 to 1 odds. We can be 99.2% certain that Hellickson’s true mean BAbip is lower than the average. It doesn’t matter if Hellickson’s BAbip this year is essentially back to normal–even when you factor it in, it’s almost impossible that the variation in his BAbip from the .292 average can be explained by chance alone.
.224, .264, .301. If a statistician sees those numbers out of context, it looks like three values from a normal distribution with a mean of .264 and a standard deviation of .04. And from a statistical standpoint, that may explain exactly what’s going on. Jeremy Hellickson’s true batting average on balls in play should be well below the league average, but there is plenty of variation involved. Sometimes he’ll have the best BAbip in baseball, sometimes he’ll be just better-than-average, and other times he’ll actually end up right around the average. But at the end of the day, Hellickson’s BAbip is going to be quite a bit better than average considerably more often than not. In that aforementioned normal distribution with a mean of .264 and a standard deviation of .04, the chances of his BAbip being greater than .300 are approximately .19, basically 1 in 5, in a large sample. Every few years, Hellickson will have some bad luck and have a season where his BAbip ends up around league average. 2013 is that year for him. On the whole, however, his BAbip should consistently be lower and that should play a big role in his success moving forward as a pitcher for the Rays.