BABIP is an overrated stat

By Jeff

Advanced statistics are beginning to take over the sports world. Right now, they are used in baseball more than the rest of the major sports. Some of these advanced stats are pretty interesting and make sense. Others  don't factor in some very important factors and need to be viewed with skepticism. One of those stats is Batting Average of Balls In Play (BABIP).

Here is a great column from ESPN's Peter Keating (ESPN Insiders only) explaining how BABIP does not always mean a player is getting lucky or unlucky. Since most people aren't ESPN Insiders (I wouldn't if it wasn't free), I've pasted the column below.


Mike's life might come crashing down upon reading Keating's column.


Otis Redding - Satisfaction

Smart baseball fans know about batting average on balls in play, or BABIP. Smarter fans should forget everything they've ever learned about it -- and I'm going to tell you why.

One of modern sabermetrics' fundamental discoveries is that pitchers have much less control over balls put into the field of play than they do over strikeouts, walks and home runs allowed. This has led to a whole new way of looking at pitchers, beginning with defense-independent pitching statistics. It has also led many pseudo-sophisticated analysts to mistakenly assume that any batter with a BABIP higher than league average must be lucky and therefore likely to decline. And that, in turn, has added to the mainstream backlash against using advanced stats at all.

Last month, for instance, Mets announcer Gary Cohen, normally an astute observer of the game, said, "David [Wright] has struck out a ton this year. Struck out 91 times. And yet he's hitting well over .300. Now, one of the stats that the sabermetrics people like to throw at you is batting average on balls in play. And if you have a particularly high batting average on balls in play, they like to think that … it shouldn't be that high, which means you're having a fortunate year and you'll come back down again. … To me that doesn't make any sense. Certain guys hit the ball harder than other guys hit it. Certain pitchers induce more ground balls or more weakly hit balls than others. That's part of what you're trying to do." Ron Darling, another smart guy, agreed with Cohen: "I think that for the average hitter, to have a high average putting balls in play, it's probably because they do have some lucky hits. But certain hitters, like Wright, hit the ball hard almost all the time." (Hat-tip: Dan Lewis at Amazin' Avenue, who posted the exchange as part of his excellent essay, "It's Time To Stop Using BABIP.")

Of course, Cohen and Darling were wrong to dismiss regression to the mean (and randomness in general) -- a huge factor in any hitter's batting average that BABIP helps illuminate. But it would also be wrong to assume that there's no skill at all in generating a high BABIP. To take an obvious example, Ichiro is batting .348 on balls in play this season. Are his 37 infield hits the product of luck? Or James Loney -- should we expect him to take a step back in 2011 just because he has an above-average .309 BABIP this year? Thing is, Loney has a career BABIP of .313. Ultimately, the relationship that's most revealing is not a hitter's BABIP to the league average but a hitter's BABIP to his expected norm.

Think about it like this. Suppose I took an important test and I scored 90 out of 100 and the class average was 75, then I had the chance to retake the test in hopes of bettering my score. Should I? If you didn't know anything about my intelligence, you surely would tell me not to try again; I would be more likely to score closer to 75 than anything above 90. But if you knew for a fact that I had the intelligence to score a 95 on the test, you should urge me to take it again; in this case, moving toward the center of my personal bell curve would improve my score. The same logic applies to BABIP. As with other stats, players regress to their own means; we should only use the league average as a proxy when we don't know what individual abilities are, or when we think there are no meaningful differences among individuals.

And there are some big differences among hitters' tendencies to hit line drives, fly balls and ground balls, each of which turns into hits at a different rate. Line drives, for instance, turn into hits at a much higher rate than ground balls do, so the more line drives you hit compared with ground balls, the higher your BABIP will tend to be. Loney, for example, has hit line drives on 25.8 percent of the balls he puts in play, the highest proportion in the majors, but Aaron Hill has generated line drives on just 10.5 percent of BIP, the lowest. We should expect Loney to have a much higher BABIP than Hill -- and he does. Various other factors influence a hitter's BABIP, too: his speed, his ballpark, and the pitching and defense he faces. And numerous metricians have come up with methods to project what a hitter's BABIP should be, given his raw stats; here's a Hardball Times study that compares various estimators.

I think, though, that estimating BABIP for individual hitters is just the first step in making batted-ball statistics meaningful. If a batter is converting balls in play more or less successfully than expected, sabermetricians should say so, then express the difference in hits or batting average. Josh Hamilton has hit line drives on 21.9 percent of his balls in play this year, ground balls on 41.1 percent and fly balls on 36.9 percent. MLB batters hit .724 on line drives, .237 on ground balls and just .138 on non-homer fly balls. Put those two sets of numbers together, and we would expect Hamilton to hit .307 on balls in play, although he actually has hit a whopping .395. But what does a .307 or .395 BABIP mean to most fans? In my opinion, the right, fan-friendly way to say this is that Hamilton has 32.8 hits more than expected in 2010. Or that his expected batting average is .294, 65 points lower than the .359 he is currently hitting. And that both differences are the biggest in MLB.

We can also use batted-ball stats to calculate "true" batting champions: According to the method I've used for Hamilton, the expected batting average leaders are Joe Mauer (.304) in the AL and Albert Pujols (.300) in the NL. And we can extend them to figure out estimated on-base percentage and slugging percentage, as this FanGraphs contributor did for Mark Kotsay. Ron Shandler of Baseball HQ has been presenting a version of xBA, or expected batting average, including a BABIP component, for years.

There are a lot of ways to go with BABIP. But the bottom line is that it's time to stop talking about BAs on BIP in relation to mythical averages and start translating them into meaningful, accessible expected statistics.

UPDATE: So, last week I showed that Trevor Hoffman gets disastrously fatigued when he's even a little overworked, and what does Ken Macha do? He uses Hoffman on back-to-back nights. Predictable result: a blown save on Tuesday night against the Reds, and another loss, instead of save No. 600. Meanwhile, the Twins, whose Matt Capps ranked first on our list of fatigued closers, went out and acquired well-rested Brian Fuentes. Brainpower isn't the only difference between winning and losing organizations, but it's key.