When I first wrote about year-to-year BABIP variation, a handful of people asked me what the "average" variation was. It's a good question--given how much batting average varies, it stands to reason that the closely related BABIP--batting average on balls in play--would flucuate even more. After all, the two components of BA missing from BABIP are strikeouts and home runs, stats that tend to be relatively constant from year to year.
Then, when I began to look at players who had "down" BABIP years and noticed that many of them didn't rebound, that raised the question of when, and by how much, BABIP declines.
The method to my carefully-controlled sanity
Here's what I did to take a stab at answering that question. First, I took every player who got at least 200 ABs in consecutive seasons between 1995 and 2005. (If you're interested in the nitty-gritty of methodology, Tango's article on aging patterns explains one problem with this approach.) I categorized those pairs of consecutive seasons by the age of the player in second year. I park-adjusted the stats, but I didn't adjust for season. (That lack of adjustment shouldn't have much impact on any of the numbers here, though it might bring the standard deviations down a bit.)
Still with me? I found the BABIP for each player for the pair of seasons included, and calculated the difference. I then grouped all of the season-pairs by age and did the rest of the work on each age group. Rather crudely, I threw everybody 37 and older into the same bucket as the sample size for each age group is perilously small at that point.
So, the second column is the key here: it's the average change that a player sees from one year to the next. In the age 24 row, that means the average player who got 200 ABs in their age 23 and 24 season saw their BABIP go down .0033 between those two seasons. The third column is the standard deviation of those changes--these numbers reflect the huge fluctuation in BABIP from year to year.
Unless you're really into this stuff, I suggest you skip down to the table.
The fourth column, "Avg fluc", directly answers one of the questions I mentioned at the outset. It tells you, at each age, how much the average player's BABIP changes, whether up or down. Age isn't too important here--the number is fairly constant. The fifth column is the standard deviation of that measure of fluctuation.
The final column is the number of players used to make the calculations for that age group. 100 or so players doesn't inspire a huge amount of confidence, but the only real solutions to that problem are greater evils: delving into minor league stats (and bringing in unreliable MLEs) or extending the study backwards and losing some applicability to the way the game is played now. At long last, here are the numbers:
|Age||Avg chg||Chg SD||Avg fluc||Fluc SD||Sample|
We know you're a nerd, Jeff, now tell us something we DON'T already know.
The most striking thing about this data is that at every age but three, BABIP goes down. (Really, the +.0005 at age 27 shouldn't count, either.) The increase at age 25 makes some sense, as players get settled in the major leagues and make better contact. The increase at age 35...well, let's just say I thought at first that was a bug in the program. I can't explain that one. Maybe players who stick around to age 34 do okay for another couple years, but that's just a blind guess.
It's commonly acknowledged that one of the major factors influencing a player's BABIP is their speed. In just about every case, a player's speed decreases (it make stay constant, but it very, very rarely increases) as he ages--thus, BABIP should decline as well. It also seems reasonable to say that a player's reflexes are never better than when they're young. While a player might improve their hitting skills, especially early in their career (hence that boost at age 25), their reflexes--which to some extent determine the quality of contact they make--will not.
I skimmed the whole thing. What's the point?
In other words, there are intuitive physical reasons why players will not improve their BABIP over time. Of course, if this study shows one thing, it's that BABIP is a giant crapshoot from year to year, so drawing conclusions from this dataset may just be a fool's errand. But if you're like me and you'd like to draw something from all the numbers, though cautiously, it seems like your favorite veteran's BABIP will probably not go up next year. However, I have good news--that overpaid vet on the other team will most likely see his BABIP decrease, as well.