DIPS for Hitters

Last month I explained the concept behind Defense Independant Pitching Statistics (DIPS) and showed that the correlation between a pitcher's batting average on balls in play (BABIP) is near zero, indicating that how often a batter gets a hit when making contact (other than a homerun) is not under the control of the pitcher.

But what about the reverse? Does a batter have any more control over whether a ball put in play becomes a hit? To answer that question I took at look at the 244 players who garnered over 200 at bats in both 2003 and 2004. I then calculated their batting average, BABIP = (H-HR)/(AB-HR-SO), and the difference between the two. I then ran a regression and found the following correlation coefficients:

BA .337
BABIP .253
DIFF .609

STDDEVP(DIFF) = .018
AVG(BABIP) = .304
AVG(BA) = .272
MAX(-DIFF) = Luis Terrero (-.109)
MAX(+DIFF) = Barry Bonds (.048)

What this indicates is that a hitter's batting average is positively correlated from year to year, a fact common sense and observation would indicate. The positive correlation of BABIP also indicates that hitters have more control over whether balls they put in play become hits than do pitchers. In my previous article I showed the correlation for pitchers was around .087. Further it is clear that the difference between BA and BABIP is around 20 points.

What is most interesting, however, is that the difference between the two has the strongest correlation at .609. While this might seem surprising at first it makes sense. Hitters with lots of power will tend to have a higher BABIP than batting average since each homerun raises their batting average while not affecting their BABIP. As a result, hitters whose hits include higher percentages of homeruns and a modicum of strikeouts will tend to have a large difference from year to year thereby increasing the correlation. By the same token hitters who hit few homeruns and don't strike out much will tend to have similar BA and BABIP and consequently will have consistently small differences in the two. At the other end of the spectrum hitters who tend to strike out alot will consistently have higher BABIP than they do BA since strikeouts make up a larger percentage of their at bats.

So from this we can conclude that it is not purely luck that a player like Barry Bonds has a 48 point difference in his batting average of .362 and his BABIP of .314, or find it surprising that among those with big differences between the two include Albert Pujols, Tony Bautista, Aramis Ramirez, Mark Bellhorn, Jason Varitek, and Moises Alou.

Even given that there is some reasonable correlation going on it is possible to find players who seemingly ran into some good or bad luck in terms of getting hits on balls in play in 2004. For example consider these two groups of players:

                 BABIP 2003   BABIP 2004
Erubiel Durazo     .287         .369
Jason Varitek      .306         .373
J.T. Snow          .307         .369
Jack Wilson        .282         .333
Darin Erstad       .285         .335
Travis Hafner      .306         .356
Damien Miller      .290         .329
Ivan Rodriguez     .337         .376
Carlos Guillen     .315         .352


Scott Posednik     .362         .275
Corey Koskie       .360         .276
Jacque Jones       .356         .283
A.J. Pierzynski    .335         .270
Brad Wilkerson     .352         .294
Tike Redman        .349         .298
Jose Guillen       .362         .312
Bill Mueller       .355         .305
Mike Lieberthal    .335         .287

In the first group the players had roughly average BABIP in 2003 and very good BABIP in 2004. It's no surprise that each of the players in the first list was considered to have a good year in 2004. In the second list are players with high BABIP in 2003 and merely average BABIP in 2004. Likewise, each of these players was considered a bit of disappointment in 2004. The interpretation of this data is that the players in the first list were a bit lucky in 2004 while the players in the second list were lucky in 2004.

The end result is that if I were a general manager I'd be a bit wary of players with really high BABIP (over .350) when they haven't historically performed at the same level (I would exclude a player like Melvin Mora who had a BABIP of .374 in 2004 almost duplicating his .364 in 2003). The players in the first list above are the ones that stand to come down to earth a bit in 2005 just as those in the second list did in 2004.