I feel like it’s about that time of year to talk about BABIP again, Batting Average on Balls In Play. There are a number of things that can happen in any given plate appearance: a strike out, a walk, a homerun, or a ball in play. A ball in play is simply a fair ball (or foul ball that is caught), balls that fielders had a chance to make a play on. Obviously some of these balls end up being hits, while some are turned into outs by the pitcher’s defense. Why should we care about BABIP? Because it is strongly correlated to pitcher success (runs allowed). Obviously, the more often that a pitcher gives up hits on balls in play, the more runs are being scored against him.
It would seem like certain pitchers would be able to limit the hits per ball in play (BABIP) against them. By jamming hitters, inducing weak grounders and popups, and generally being able to get hitters to hit to his better fielders, a pitcher should be able to reduce his own BABIP, and thus ERA, right? Well, in a word, no.
Over the last 20 seasons, starters* have accumulated an average BABIP of 0.294, and 0.296 in the last 10 seasons, so this gives us a good idea of the league average. In Greg Maddux’s 1998 season, he sported a 0.267 BABIP. At the the time, one might have argued it was due to his brilliant scouting and knowledge of batters’ weak points, heroically inducing weak contact in the clutch en route to a 2.22 ERA and 18 wins. But in the 1999 campaign, his 0.331 BABIP would have had us believing the exact opposite (only a year later), that he was actually bad at allowing hits per balls in play. If BABIP were truly controlled by the pitcher, we shouldn’t expect such a large spike from season to season. Not surprisingly, Maddux gave up 57 more hits in 1999 than he had the year before, and his ERA rose from 2.22 to 3.57. You better believe BABIP had a lot to do with that ERA spike.
This season, when Ubaldo Jimenez of the Rockies was sitting pretty with a 0.78 ERA after 11 starts, casual observers may have predicted 30 wins and an ERA under 1.00. Well, it’s been 13 starts since, and his ERA over those last 13 has been, wait for it, 4.34! What gives? Jimenez’s BABIP after his 11th start had fallen to ridiculously low levels at 0.228. We saw how Maddux’s 6% BABIP spike hurt his ERA, so a return to the league average (a 7% increase) for Jimenez would have been catastrophic! Oh wait, it happened. His BABIP over the last 13 starts has been a much more appropriate 0.307. Pitchers are simply not able to sustain BABIPs that stray too far from league average, a range of about 0.290 to 0.300.
Currently, Atlanta’s Tim Hudson flaunts a microscopic 2.15 ERA, accompanied by a suspicious 0.241 BABIP. If history is any indicator, both his BABIP and ERA should start to rise at approximately 8:40 PM ET on August 23, 2010, his next start. Though a more sound prediction would be that over his remaining starts this season, we can expect his ERA to hang around the 4.00 range.
“But Matthias!” you scream, “you can’t just take two examples and make a case. There are hundreds of pitchers who pitch every season!” Good point, my friend, good point.
Let’s take a little peek at the 2,750 pitchers’ seasons of at least 100 innings over the last 20 years. Good pitchers should have lower BABIPs, right? Pitchers that have good control (low walk totals), strike out a lot of batters, and give up weaker contact must have lower BABIPs, right? The only indicator we have of weaker contact would be homers allowed per fly ball (HR/FB). If pitchers are giving up weak contact, we’re likely to see it in this statistic.
Remember that the BABIP average of the entire group over the last 20 seasons is about 0.294. If we trim down our sample to only the pitchers with above-average strikeout rates, below-average walk rates, and those who give up the lowest HR/FB rates, we really ought to have a group of hurlers who can induce weak contact and limit hits on balls in play.
Of those 2,750 pitchers, I took only the ones with K/9 above 8.0, BB/9 below 3.0, and HR/FB below 7.0%. These figures guarantee that we’re analyzing pitchers who are better-than-average in all three stats. The list was whittled down to 71 pitchers’ seasons matching the above description. Among them were both Tim Lincecum and Zach Greinke’s 2009 Cy Young seasons, Jake Peavy’s 2007 Cy Young season, Pedro Martinez’ 1999 Cy Young, John Smoltz in 1996, and Roger Clemens in both 1997 and 1991…you get the idea this is a quality group of pitchers. If anyone can limit hits on balls in play, it’s these guys.
Well, the average BABIP of these 71 pitchers was 0.295. One tiny point higher than the entire sample. The take away? Pitchers just can’t control their BABIPs significantly. So BABIPs that find themselves too far from the 0.290 to 0.300 range are due to regress any day now, taking with them the pitcher’s ERA. I’m looking at you, Tim Hudson.