Mailbag Estimates

Jason Michael Barker

Leading off, thanks to everyone who wrote in regarding Thursday's column on estimating ERA. I got more mail about this topic than I have about any thing else I've written, which hopefully means our readership is growing. Either that or you were just particularly upset with me on this occasion.

Looking back, it seems I was a bit too hasty in chalking up differences between estimated and actual ERAs to bullpen performance alone, and you didn't hesitate to let me know. Further, several other good ideas were put forth as possible explanations for the error of the estimates. So without further ado, let's get on to some of your mail.

We begin with Tony, who writes:

I have noticed that certain hurlers like Maddux and Martinez seem to have the ability to "scatter" hits instead of getting rocked around the park. You can watch a game where Martinez will give up ten hits, allow one run and win, while his opposing pitcher, say Eric Milton, will allow six runs on five hits and lose. Is there a skill in stranding baserunners or is it luck?

This is certainly a good point. It's much better, for example, to give up a single each inning for nine innings than to allow all nine hits consecutively, and simply looking at a pitcher's on-base and slugging percentages allowed doesn't tell us a thing about when he allowed the damage.

As we'll see as we move on, this sort of thing is typical of the explanations offered for the difference between actual and estimated ERA -- there are a variety of factors that just aren't represented in OBP and SLG.

For more on the possibility of "clutch pitching," we come to Jon:

Here's a little idea... This is based upon my knowledge of Pedro Martinez, being a big Sox fan and all. One thing about Pedro that I noticed this season is that he really bears down and becomes even tougher when he gets a guy on base. He becomes even more focused and I've seen his nasty stuff get nasty beyond description (shouldn't be shown to those who are weak-hearted) when the occasional runner gets to third base.

Another good explanation, and yet another thing which doesn't show up in OBP and SLG. Looking at Martinez specifically, Jon's observations are right on the money: Pedro allowed a stellar .255/.312 (OBP/SLG) with the bases empty, but was even tougher with runners on to the tune of .235/.249. Wow!

Moving on to my hometown team, I observed a similar pattern of pitching from Seattle's Freddy Garcia last season. With the bases empty, Garcia allowed a .378 OBP and .447 SLG. With such mediocre numbers, you might wonder how he was so successful in 1999, right? Wonder no more, because with runners on base, Garcia held hitters to a .309 OBP and .346 SLG. With runners in scoring position, those numbers drop to .310/.317.

Wayland, who sent me a variety of statistics (overall, none on, runners on, runners in scoring position) for Juan Guzman, Brad Radke, Eric Milton and Steve Trachsel, further elaborated upon this point. Guzman and Radke were the "luckiest" pitchers of 1999, while Milton and Traschel were the two hurlers at the opposite end of the spectrum.

To make a long story short, Guzman, much like Martinez and Garcia, was much harder on opposing hitters with runners on base. Radke's numbers were a bit more ambiguous: he did cut down on opponents' batting average and OBP with runners on, but also allowed a slightly higher SLG. Apparently the added power didn't hurt him, though.

Milton is a very extreme case of falling apart with runners on: he allowed .299/.406 overall and .278/.353 with none on, but allowed .335/.497 with runners on and .348/.563(!) with runners in scoring position. If Eric Milton is to become an elite pitcher in the coming years, and I think the ability is there, he'll have to maintain his focus with runners on base.

Steve Trachsel's stats were also quite ambiguous, as there wasn't much variation between his numbers with runners on and the bases empty. At this point we have to look for other possible explanations, which is once again where you come in. Oh, and thanks to Wayland for all the numbers.

One idea comes to us from Seth, who writes:

I read your article about the aERA and eERA. Seems to me that if you are adding SLG and OBP, which really translates to whether there are men on base and where, the most important factor will be strikeouts and infield defense. Bullpen only comes into play 1 inning per game for a starting pitcher, if that.

If, as the formula assumes, there are men on base, the pitcher who strikes out the most batters (as opposed to giving up RBI groundouts and sac flies) would seem to have the edge. It seems to hold a bit, with the Unit [Randy Johnson] and Pedro up near the top... but Guzman? And Kevin Brown is minus 5.2. I'd like to see the K/9 stats for these guys...

Strikeouts are certainly an interesting idea, and one that didn't occur to me. It does make sense, however, since strikeouts with runners on base keep the ball out of play and prevent runners from moving up or scoring on groundouts or fly balls. For what it's worth, Guzman struck out 155 hitters in 200 innings last season, or about seven per nine innings. For comparison, Kevin Brown was up close to eight.

For yet another explanation, Ray writes:

You missed the obvious: pitchers that rely on groundballs will tend to have a higher estimated ERA than actual ERA. Just look at the number of GIDPs that Mike Hampton induced last season. That was no fluke; it was a reflection of his pitching style.

Since neither OBP nor SLG account for DPs, the "OBP*SLG*.31" formula will typically over-estimate the ERA of groundball pitchers and under-estimate the ERA of flyball pitchers (see: Trachsel, Steve).

Mike Hampton had a 2.56:1 ground ball to fly ball ratio last season and induced 38 double plays last season, both good for second in the majors behind Baltimore's Scott Erickson. And have we finally found the explanation for Trachsel's horrible performance relative to his estimated ERA? There isn't room here to take an in-depth look at the relationship between a pitcher's G:F ratio and his estimated ERA, but it's certainly an area worth looking into.

Finally we come to Daniel, who presents a grab-bag explanation:

Aside from having good relievers who retire inherited runners for them, pitchers can get a statistically intangible benefit from good defense. Having a cannon-armed RFer who keeps the runner from advancing on a sacrifice, a double-play machine at 2B and short, a catcher who doesn't allow passed balls and who prevents wild pitches -- these are some more factors that keep real earned runs from scoring but don't factor into the eERA.

And there you have it. There are numerous factors influencing ERA that don't show up in a pitcher's OBP and SLG Allowed, and Daniel has hit on several of them we have yet to mention here.

Thanks again to everyone who wrote in regarding estimated ERA; I've had a good time putting this particular column together. For a final thought, I'd like to pass along another comment from a reader email: The reason this particular statistic is interesting is that it invites speculation, as opposed to crushing it.

about the author

Jason Michael Barker is attempting to augment his income as a baseball chiropractor. Recommend Hideo Nomo as his first potential client at jmb@strikethree.com.