Saturday, February 11, 2006
PECOTA Hates Sox Pitching...Again
Last year, I complained that PECOTA, Baseball Prospectus's projection system, was unfairly harsh in estimating the White Sox's pitching. PECOTA projected that the Sox's lowest ERA would come from El Duque, and that he would only manage an 4.35 ERA in his own right. Thankfully for the White Sox, PECOTA was wrong, horribly wrong in projecting the Sox's starters in 2005:
|Pitcher||2005 PECOTA||2005 Actual|
PECOTA was high by more than one run for Buehrle, Garland, and Contreras, two-thirds of a run for Garcia, and low by two-thirds of a run for El Duque.
This year, PECOTA is once again projecting dire straits for the Sox pitching staff, with everyone but Vazquez coming in over a 4 ERA:
(FYI, Brandon McCarthy is projected at a 4.43 ERA). All of these projections - save Vazquez's - would be poor seasons for the Sox starters.
Using my clever cover of "George in DC", I queried Nate Silver about these projections on Baseball Prospectus's most recent chat. My first attempt:
George (DC): Nate - why does PECOTA hate the White Sox
pitchers again? Vazquez is the only guy below a 4 ERA? He's the 5th starter on
this team! Last year it got them very wrong as well. What's the bias there? Does
it think that US Cellular is just too tough to have any kind of sustained
success. Even Buehrle looks average in PECOTA.
Nate Silver: George, I buy into the projections for the White Sox pitching staff, with the possible exception of Contreras and maybe Bobby Jenks, each of whom have a lot of
backstory going on that PECOTA won't be able to pick up on. There's really three
different things going on that harms the White Sox pitchers:
1) Most of them performed significantly better in 2005 than in 2003 or 2004.
2) As you mention, it's an uphill battle to put up good ERAs in The Cell.
3) The White Sox got something like +50 runs from their defense last year, all of which finds its way into the ERAs of individual pitchers. We're projecting that their defense this year will be much closer to league average. Part of that is because they punted
Aaron Rowand, but part of it is that a lot of guys like Konerko and Podsednik
posted defensive numbers that were out of line with their past histories.
OK, so I can understand regression to the mean as a concept for 2006. They had a heck of a year in 2005, and you might expect some drop off. But a run per nine innings for most of them?
The second reason I myself put forward, but note that Buehrle has been putting up sub-4 ERA's at the Cell since 2001 (with one year as an exception).
The third reason did not fly with me, however. First, I think that Brian Anderson is going to be 90-100% of the defensive centerfielder Aaron Rowand was. And, as I point out in my rejoinder, Podsednik is not going to regress:
George (DC): Nate In response...weren't the real improvements
from Garland and Contreras in 2005? And given Garland's 2005 was his age-26
season, isn't his improvement more likely to stick? Also, Podsednik moved
from CF to LF last year, so I'm not surprised that his numbers moved up
considerably. I wouldn't count on much of a dropoff in 2006 as he continues
to play a position once manned by Carlos Lee (Horsehands, as he was known to
Sox fans). Konerko was only about 10 runs better than average, so his
defense isn't going to make THAT much of a difference. I'm just surprised by
the pessimism for 2006, when PECOTA was so far off in 2005 (Buehrle 4.47,
Garcia 4.55, Garland 5.05, Contreras 4.91, El Duque 4.35).
Nate Silver: George, I've said that I think Contreras' forecast is low. Garland's
improvement in 2005 was the result of a fairly ordinary maturation process -
he got better at figuring out which pitches he should throw on which counts,
and things like that. He didn't develop a new delivery or a new pitch or
something like that, and he still has some trouble missing bats. I think the
comparison to Jeff Weaver is instuctive (although he has an advantage
over Weaver in the durability department).
So, bottom line is that I'm not going to get an answer from Nate on PECOTA and the Sox. We'll see if there is some sort of statistical bias there this year.
On a happy note, PECOTA likes Sox rookie Brian Anderson to have a nice season - hitting .269 with a .329 OBP and a .468 SLG. That's a bit better than Aaron Rowand's .270 / .329 / .407 line in 2005. For $3 million cheaper, I'd say the Sox won't be any worse off with Anderson patrolling center if that prediction bears out.
It's interesting to hear the BP guys talk about exactly what PECTOA is looking at. I surmised that it would indeed factor things like park and DIPS into the projections, and also the 3 year means.
One thing I can think of for Buehrle is that his peripherals consistently undervalue his worth. BP's PERA has been much higher than Buehrles ERA each of the past 3 seasons.
I don't think Buehrle will post a 3.12 ERA again. He got a bit lucky in that only 7% of fly balls against him turned into homeruns.
I certainly project him under four though. His W/PA has gotten better each of the last three years and I don't see him losing the accuracy that makes him so deadly.