A Few Thoughts on BABIP

Last week, I wrote about how the dramatic turnaround Josh Beckett has experienced this year has been mostly driven by a huge change in his BABIP, and in doing so, noted that this year’s version of Beckett doesn’t seem to be that different from last year’s version. The always insightful David Pinto responded, using Pitch F/x data and heat maps to show that Beckett’s pitches are showing a real difference this year. His conclusion:

To sum up, Beckett exhibited less control of a straighter fastball in 2010. Batters hit that pitch harder. Beckett’s bad luck seemed more due to an injury hurting his mechanics than balls finding holes on good pitches.

DIPS is often right, as it was on Dan Haren. In the case of Beckett, however, there is reason to believe that his improvement is more than just regression to and past the mean. Sometimes pitchers make their own luck.

In reality, I don’t think David and I actually disagree here — he just corrected some sloppy writing on my part, and that got me thinking that I probably needed to talk more about BABIP and regression, because too often, we just sum up variation as luck but don’t explain what we really mean by that. So, here’s a general take on what I see as the main causes of variations in BABIP.

Good or Bad Defense

I’m not overly surprised that the Brewers have the fourth-highest BABIP allowed in the Majors this year, as we all knew going into the season that they were gambling that they could win with a lousy defensive roster. Between Yuniesky Betancourt, Ryan Braun, and Prince Fielder, they have three guys who are among the worst defenders at their position in all of baseball – Carlos Gomez and Nyjer Morgan can only cover so many of their teammates’ sins.

This isn’t luck — this is just non-pitching skill that affects the results of a ball in play. This area has been extensively covered the last few years, and is generally accepted as a variable that needs to be accounted for, so we won’t spend too much time here.

Good or Bad Luck

Last week, Ervin Santana was trying to hold on to a 1-0 lead in Seattle. With two outs and the bases loaded, he got Carlos Peguero to hit a weak ground ball up the middle. Erick Aybar was in perfect position to pick up the ball and throw Peguero out to end the inning, but on it’s way to his glove, the ball hit the base and bounced into left field.

There’s no other way to describe that play than luck (good or bad, depending on your perspective). The pitcher got weak contact hit directly at a fielder, but because the ball bounced just so and deflected off an object embedded in the ground, it ended up as a two-run single. Not all lucky plays are this obvious, but there is no denying that outcomes like this happen, and they affect a pitcher’s results. Balls hit down the line are often fair or foul by a manner of inches, and the results are drastically different based on minuscule variations in when the batter started his swing. Balls are hit on a line but right at a defender. A wind gust knocks down a fly ball to the alley, or two defenders forget to communicate and a ball falls between them. This stuff happens, and the results obviously don’t reflect on the quality of pitch thrown.

These are the kinds of plays that most people think of when we describe a pitcher as being lucky or unlucky. And, given their frequency, it’s easy to understand their skepticism when someone states that a pitcher has been having a lot of luck (either good or bad), because these don’t happen so often that it’s easily memorable in most cases. These plays do have an impact on a pitcher’s BABIP, but most of the time, they’re not the driver of huge swings one way or another for any one pitcher.

Good or Bad Pitching

This is what David Pinto is talking about in his post about Josh Beckett, and it’s something I should have been more clear about. As many Boston fans noted, the Beckett they saw last year wasn’t just a victim of bad defense (though that may have been part of it) or bad luck (possible, but we don’t know for sure), but was giving up a ton of hits because he was throwing pitches that were just getting whacked. And this is the area where most fans have a hard time accepting DIPS theory.

When they see a guy throw a belt-high fastball and it gets ripped into the gap for a double, it sure doesn’t seem like bad luck; it seems like the guy threw a terrible pitch and got punished for it. And, if he repeatedly throws pitches that are getting hit hard all over the field, it’s tough to reconcile what just happened with a theory that says it won’t keep happening because the pitcher just got “unlucky.”

As David shows with his heat maps, and as most fans understand from just watching the sport for a while, not every pitch is equally hittable. It is certainly possible that Beckett just threw more hittable pitches last year than he is this year, and I should have been clearer about that in my post last week.

However, there is a commonality between all three of these areas that helps stats that just treat all BABIP variation as luck actually work pretty well — whether it’s defense, luck, or quality of pitches, all of them are mostly unsustainable in the long run for most pitchers.

Exceptionally bad defenses don’t stay together for too long. Pretty soon, the Brewers will realize that Yuniesky Betancourt is horrible, and he will be replaced. Even for a pitcher on a team with great defensive teammates, injuries, trades, or free agency can rip that apart pretty quickly. Obviously, a pitcher who has been getting lucky with things like sun balls or line drives right at fielders will see that come to an end just due to random variation.

But there’s also an unsustainable effect in play when it comes to throwing pitches that are easy or hard to hit. If a pitcher throws a bunch of hanging curve balls that are getting smacked all over the field, the catcher will eventually stop calling for them. Or, he might learn that he can’t throw that pitch in that location, and make an adjustment. Hitters make adjustments, too — they’ll watch video, realize that a pitcher likes to spot a certain pitch in a particular location, and once they expect it, the results will change.

Let’s use Charlie Morton as an example. In his first three starts of the year, he posted a .164 BABIP by throwing 90% sinkers to one particular spot, a massive change from the approach he had used last year. Hitters were hitting weak ground balls right at defenders, and to anyone who watched him pitch, it certainly didn’t look like good luck or good defense. But, it was unsustainable all the same — hitters began to adjust, the scouting report got out around the league, and now Morton has a .322 BABIP after getting lit up again last night.

A stat like FIP doesn’t distinguish between whether variations in BABIP are due to defense, luck, or the quality of pitches thrown, and this is where we need to do a better job in specifying which factors we think are actually causing the variation. In the case of Beckett, I shouldn’t have just lumped everything under luck — he probably was throwing pitches that were just easier to whack last year.

It doesn’t necessarily change the conclusion in most cases — with the exception of in-season defense for a player that isn’t changing teams — because whether luck or not, most BABIP variations are not something that predict future BABIP all that well, and for most of our purposes, that’s what we’re really trying to do. However, David is right to point out that pitchers can create their own luck, and I should have done a better job of clarifying what the cause of Beckett’s change in BABIP actually was.

It doesn’t mean I think he can keep posting a .220 mark this year, but it’s probably not accurate to say that it’s a result of balls finding gloves with regularity or other things that would generally be considered luck. Beckett probably has thrown a lot of good pitches this year, generated weak contact, and looked like a tremendous pitcher in the process. History suggests that this won’t last, but that’s different than saying it hasn’t happened already.