My Little Pony? Isn’t that a little… err, GLBish?

In limited sample sizes BABIP numbers are subject to large fluctuations that cause pitchers to have extremely misleading ERAs at the beginning of the season.

If a pitcher in 50 IP has a BABIP around .200 you can expect that he’s due for a fairly large regression ( for an example of this in action, see Elmer Dessens’ 2002 season vs. his 2003 season). If a pitcher has a BABIP in the .400 range they could be going through a string of bad luck.

However, when evaluating pitchers at the high end of BABIP numbers, it’s also instructive to look at their component numbers like ground ball, fly ball and line drive rate. Sometimes a high BABIP can be the result of a large numbers of fly balls and ground balls finding holes. But other times it can be an indicator that a pitcher has simply “lost it”. If the numbers show that they have a high BABIP that’s largely due to an extremely high line-drive rate, it likely means that the pitcher is toast. To be sure, one should also look at strikeout, walk and home run rates.

But when you combine all these small samples into one larger total innings sample (like DER), the random fluctuations tend to even themselves out.

Jim, feel free to copy and repost this so that Foul will read it.

since one is based on the other. Unearned runs are based on errors as well, so you’re basically judging defense entirely on errors. Not the most instructive/objective stat.

F% and E are basically the same stat, so if you make a lot of errors, your fielding percentage will be poor. Unearned runs are a bit like the RBI of fielding statistics, because they depend almost entirely on the situation when they happen.

If an outfielder drops a fly-ball with the bases empty and two outs, there’ll be a runner on second with two outs,and that will likely be the end of it. If exactly the same thing happens with the bases loaded and no outs, you’ll see at least three unearned runs. DefEff doesn’t play favorites like that.

As for why we rate okay in DefEff but not in errors, I mentioned it before, but I think we make more difficult plays than other teams, and bungle more easy ones. The former don’t count as errors for the teams that fail to make them, but the latter do go against our record. Overall, these balance out and we make about an average percentage of outs.

If you look at just this year’s worth of data, you are going to see a very strong correlation between low BABIP and low ERA. This is part of the random variation we’re talking about. Look at Elmer Dessens -

2002 – .268 BABIP 3.03 ERA
2003 – .330 BABIP 5.04 ERA

his career BABIP was around .309. It’s fairly clear that in 2002, his low ERA was largely due to luck.

If you look at just this seasons worth of BABIP numbers, you’ll see that the ERA leaders all have extremely low BABIP numbers. While some of this is skill, a lot of this is just luck. Yes, there is the ability by some pitchers, like extreme groundballers to have better BABIP than normal, but the true value difference is more a .290 true average compared to a league average of .300. Haren’s BABIP is .271 this season, and Webb’s is .261. This would indicate two things, they’ve been somewhat hit lucky and they’re pitching in front of a good defense.

The trends in BABIP numbers become clearer when you look at a larger sample size. Career BABIP for Arizona pitchers -

Webb – .288
Haren – .291
Owings – .282
RJ – .294
Davis – .310
Ed it’s gone – .323
Med it’s gone – .280
Lyon – .302
Qualls – .292
Pena – .260
Cruz – .298
Slaten – .275

In some ways, Edgar’s BABIP is misleading. In ‘06 and ‘07 he put up BABIP of .290 and .286 respectively. His numbers are brought up by his disastrous ‘03 and ‘04 numbers (.417 and .370). This year he’s also been hit hard, but I’d have to look at the component data to see if he’s just going through bad luck or if he’s no longer fooling anyone.

But going back to the numbers, on average, a pitcher should have a BABIP around .300. As you can see, the most that good pitchers tend to influence the numbers is around 10%. In small samples, they can put up numbers even lower than this, but not for anything longer than a season or so. If one team had a rotation full of Brandon Webbs and another team had five edgar gonzalezes you would likely see large variences in DER, but in the real world I think that these differences prove statistically minor.

Also, to go further in depth on the high BABIP discussion, I repost:

when evaluating pitchers at the high end of BABIP numbers, it’s also instructive to look at their component numbers like ground ball, fly ball and line drive rate. Sometimes a high BABIP can be the result of a large numbers of fly balls and ground balls finding holes. But other times it can be an indicator that a pitcher has simply "lost it". If the numbers show that they have a high BABIP that’s largely due to an extremely high line-drive rate, it likely means that the pitcher is toast. To be sure, one should also look at strikeout, walk and home run rates.

I haven’t yet looked at EdGon’s component statistics, but if he’s showing an extremely high line drive rate, there’s definitely something to be worried about.

And, as dahlian has pointed out, Webb does have a better career BABIP than most. However, he is an outlier for obvious reasons related to his sinker [there are other, rare pitchers who can affect BABIP – such as knuckleballers like Wakefield] . Most pitchers simply do not have the ability to conjure up weak groundballs like Webb. Indeed, because of that, good infield defense behind him is probably less important, since the balls are easier to field. They’re busy, but not on particularly difficult plays,