Ballpark Estimates, IIJason Michael Barker

Last Thursday, I took a look at Bill James' formula for estimating a pitcher's ERA based on his on-base and slugging percentages allowed, namely eERA = OBP*SLG*31. Overall, the estimates were pretty good, but there were a number of pitchers who severely over- or underperformed their estimate, leading to a debate over why that might be. I suggested the performance of the starter's bullpen, while you wrote in with a variety of possibilities.

Today I'd like to delve deeper into this topic by looking at the performance of a few pitchers over the past several seasons. I can hear you now: "Three consecutive columns on the same topic?" Normally I wouldn't devote so much time to this, but there are two forces at work here. One, I think this is a pretty interesting discussion, and two, there's nothing else going on in MLB right now.

For this study, I looked at pitchers who have thrown at least 200 innings each of the past four seasons (or since 1996). I used 1996 as the cutoff in order to eliminate the slightly shortened 1995 season. As it turns out, there are only eleven such pitchers: Kevin Brown, Tom Glavine, Pedro Astacio, David Wells, Scott Erickson, Greg Maddux, Brad Radke, Pedro Martinez, Steve Trachsel, Mike Mussina, and Charles Nagy. (By the way, if anyone can guess how that list is ordered, drop me an e-mail and you'll get a hearty "congratulations" for your trouble).

There isn't room to show you all eleven charts, but here's an example of what I did for each pitcher:

Mike Mussina
Year	OBP	SLG	eERA	aERA	Difference	G:F Ratio	Extra Bases	Runners Erased
1996	.325	.440	4.43	4.81	-7.9%	1.18	17	28
1997	.282	.388	3.39	3.20	5.9%	1.39	19	22
1998	.283	.383	3.36	3.49	-3.72%	1.46	39	23
1999	.312	.411	3.98	3.50	13.7%	1.55	22	28
Career	.293	.392	3.56	3.50	1.7%	0.98

The year, OBP and SLG columns are self-explanatory. eERA, if you recall, is the pitcher's estimated ERA based on the formula. aERA is his actual ERA, and Difference is how far off the estimate was, as a percentage of actual ERA.

There are also several new additions to the chart. G:F Ratio is the pitcher's ground ball-to-fly ball ratio, which we identified as a factor influencing aERA without being represented OBP and SLG (ground balls induce more double plays). Extra Bases are yet another factor not represented, and are computed by adding sacrifice flies, sacrifice hits, wild pitches, balks, and stolen bases -- all things which give runners additional bases.

Similarly, Runners Erased combines double plays and failed stolen-base attempts, both of which remove runners without being represented in OBP or SLG.

I chose Mussina because his numbers go along with what seems to make sense intuitively, and in general they follow the pattern of the ten other pitchers. In 1996 he slightly underperformed his estimated ERA. That year his G:F ratio wasn't all that high, although his extra bases were relatively low.

The next year Mussina increased his G:F ratio while keeping his extra bases down, and as a result he outperformed his eERA. In 1998 he continued to improve his G:F ratio, but allowed a ton of additional bases, and his performance suffered. Finally, in 1999 he once again improved his G:F ratio and cut his extra bases at the same time, resulting in a big overperformance in terms of ERA.

You'll also notice that his career difference is quite small, which was also true of the 11 pitchers I looked at -- over time, a pitcher's "luck" seems to even out. Pedro Martinez, for example, has an actual career ERA (2.83) identical to his estimated career ERA. In fact, of the nine pitchers for whom career data is available (Greg Maddux and Kevin Brown pitched prior to 1987, before OBP and SLG were kept for pitchers), the "worst" estimate was for Tom Glavine, and even that was only 3% off.

A few thoughts on each pitcher with respect to eERA:

Kevin Brown. With the exception of last season when his G:F dropped to 2.20 (he's up around three for his career), Brown has been a repeat overachiever in the latter half of the '90s, when he was one of the better pitchers in baseball.

Tom Glavine. It's an every-other-year story for Glavine: 15.4%, 1.7%, 22.3%, 1.5%. All this despite relative consistency in his G:F ratio and extra bases allowed.

Pedro Astacio. He was clearly rattled by his first full season in Coors Field, underperforming his eERA in 1998 by a full 10%. He rebounded last season to break even. If he could ever get his G:F ratio up over his career average of 1.32, it would be a big help for the Rockies.

David Wells. A classic underachiever, Wells has missed his eERA three of the past four years, and the one year he didn't, he only managed to beat it by 1.7%. His career G:F ratio of nearly 1.0 isn't helping matters any.

Scott Erickson. Despite being one of the top ground ball pitchers in the league, Erickson hasn't overperformed his eERA as often as you might think. In fact, he's undershot it two of the last four seasons, mostly due to a ton of extra bases allowed.

Greg Maddux. Why is Maddux such a great pitcher? Because he consistently beats his eERA, of course. I don't have career numbers for him, but here are his percent differences since 1996: 1.5%, 12.3%, 8.6%, 13.2%. The Smartest Pitcher Who Ever Lived helps himself out with a career G:F of around 2.5.

Brad Radke. Radke has turned his career around in recent years by going from an extreme fly ball pitcher to a so-so ground ball pitcher. He's also is one of the best in terms of extra bases allowed, helping to compensate for his mediocre G:F ratio.

Pedro Martinez. A much lower G:F than I would have thought, but he's been an overperformer each of the past three years, probably due to his stellar pitching with runners on base. (I unfortunately was unable to find those splits for prior years, so I had to leave it out of these case studies. Despite that, it does appear to be an important factor in ERA relative to eERA).

Steve Trachsel. His best season, 1996 (3.03 aERA, or 25.4% better than his eERA), just happens to coincide with the year he had a G:F of close to two and allowed just 20 extra bases. He's dropped off in both stats since then, and has gotten shelled.

Charles Nagy. A career overperformer, mostly due to a good G:F ratio. That ratio has been in decline lately, however, so a big crash might be coming (and Nagy hasn't been all that great recently anyway).

about the author

Jason Michael Barker won't go hungry this year, what with his Y2K stockpile of Cal Ripken Crappy O's cereal. Back up his theory that the quality doesn't matter if it's for charity at jmb@strikethree.com.