There's been a lot of talk about BABIP in the comment threads recently.  I went looking for a good overview of the topic online, but came up empty.  So here I am, writing one myself.

BABIP stands for Batting Average on Balls In Play.  For the most part, that's batting average excluding home runs and strikeouts.  It's what happens in the batter-pitcher matchup excluding the "Three True Outcomes" that Rob Deer and Russell Branyan love so much.

Intuitively, BABIP is about this: The pitcher doesn't have much control over the outcome of a batted ball once it leaves the bat, unless it leaves the park.  Batters have some control, but probably not as much as you think.  A pitcher's skills can substantially be determined by looking at how he does in the Three True Outcomes (lots of K's, not so many W's and HR's), while a batter's skills are at least partly defined by them as well.

This is counterintuitive, I know.  It *seems* like good pitches lead to weakly-hit balls, and bad pitches to doubles in the gap.  That may be true for any one pitch, but in the end, it largely washes out.  Plenty of weakly-hit balls turn into hits; lots of well-hit balls end up in Mike Cameron's glove.  

In the end, it doesn't matter very much whether it makes sense or not...what follows is statistically well-established.  Don't believe the folks out there who say it's all black and white--it's not.  Pitchers and batters have some control over batted balls.  But not very much.

Luck and skill

The most important thing to understand about BABIP is that, more than almost any other baseball stat, it is dependent on luck.  That isn't to say that every single high or low BABIP is lucky or unlucky, just that variations are much more dependent on luck than, say, ERA or doubles.

How much it is dependent on luck is different for pitchers and hitters.  In this Baseball Prospectus article, Nate Silver claims (paraphrasing and simplifying somewhat) that 93% of variation in BABIP is defense and luck, while 7% is based on the pitcher's skills.  League average BABIP usually hovers around .300, so if you see a pitcher with a BABIP much higher or lower than that, odds are it isn't going to last.

Evaluating hitters

The same is true for hitters, but to a lesser extent.  Some of the best hitters in the game--think David Ortiz and Ryan Howard, for instance--hit a lot of line drives.  Certainly, a line drive is harder for the defense to turn into an out than a ground ball, a pop fly, or even a fly ball.

Now that Baseball-Reference has splits based on batted-ball types, we can look at some averages.  Here's how NL hitters did for three major batted-ball types:

• Ground Balls: .245
• Fly Balls: .134  (that includes home runs)
  
• Line Drives: .719 (including a few home runs)

(The fly ball rate would be higher if you separated out pop flies, almost none of which turn into hits.)

In other words, if you hit a lot of line drives, you'll have a higher BABIP.  And unlike BABIP for pitchers, line drive percentage is fairly consistent.  If you hit a lot of line drives this year, odds are you'll do the same next year, within reason, anyway. 

Putting all of this together, we can see better when batters are getting lucky or unlucky.  If we have a lot of time, we can take the expected BABIP for each batted ball type, see how many of each type the batter hit, and determine what his BABIP should be.  There is a shortcut, though.

League-average line drive percentage is about 18%--or about 12% lower than the average BABIP.  Since much of variation in BABIP depends on line drive percentage, we can just use this one number.  Take a player's line drive percentage, add 12%, and you have their "expected BABIP," or eBABIP.

Here are a few big-name Brewers and some of these numbers on them for 2007:

In this small sample, you can see that BABIP and eBABIP are usually pretty close together, but Braun's numbers stick out.  He was due for a downswing, and so far this year, he's getting it.  You can see line drive percentage on the Hardball Times player pages, which are linked in the left sidebar.

Back to pitchers

It's possible--though uncommon--for a hitter to have a consistent BABIP much higher than league average.  For instance, Ryan Howard's BABIPs from 04 to 07 were .375, .354, .356, and .328.  It's also possible for a pitcher to be consistently above or below average, but as I noted above, it's much less likely.

If a pitcher has an extensive track record, a high or low BABIP might be sustainable.  For instance, Tom Glavine has a better-than-average career BABIP of .285.  Incidentally, it's drifted up as he has aged, but 5 to 7 years ago, it would've been a good bet to expect Glavine to put up a .270-.275 BABIP.

For most pitchers, though, we don't have that much information.  (We do have minor league data back to 2005, which is a start.)  There are two things that can affect BABIP for those guys:

1. Ground ball tendencies
   
2. Defense

As we saw before, ground balls are more likely to turn into outs more than are balls hit in the air.  So if you have a sinker that turns into lots of grounders, you can keep your BABIP down.  Ground ball rate is one of the stats that stays the most consistent from year to year--it is even reasonably consistent from college into the pros.

Also variable is team defense.  BABIP measures how many batted balls turn into outs, and defense has a lot to do with that.  For instance, the Brewers weren't very good in the field last year, so their pitchers had a BABIP of .312, compared to league BABIP of .302. 

This was one of the reasons why some people were skeptical of bringing Suppan to Milwaukee--he left one of the best defenses in baseball for one of the worst.  And sure enough, his BABIP went up.  In 2005 and 2006, it was .296 and .298, and last year it was .324.

The abstract is at the end

If you only take a few things away from this article, here's what to remember:

• As always, sample size matters.  A week, a month, or even an entire season doesn't give you enough data to firmly establish that a player will be able to maintain a BABIP well above or below average.
  
• Most batters, and almost all pitchers, should be expected to have a BABIP around .300.  If it's higher or lower, luck may be playing a big part.
• If a pitcher gets a lot of ground balls, or is pitching in front of a good defense, he might keep his BABIP below .300.
• If a batter hits lots of line drives, he could keep his BABIP above .300.