September 27, 2006
Lies, Damned Lies
While it's evident that there is nothing like a Diamondbacks-style Randy Johnson-Curt Schilling thunder-and-lightning duo among this year’s potential playoff teams, it’s unclear just which starting staff is the best of a relatively weak lot. The Twins, with Johan Santana & the Nobodies? The A’s, with Rich Harden joining Barry Zito and Dan Haren? The Tigers, who lead the majors in run prevention?
The best way to answer the question would be to run PECOTAs for all the pitchers involved, which would help us to separate skill from luck--Chien-Ming Wang has a better ERA and a better VORP on the season than Jake Peavy, but only those folks immediately east of the Hudson River would rather have Wang start a playoff game. However, PECOTAs are time-consuming to generate, so in their absence we’ll use a favorite toy of mine. I call this toy QuikERA (QERA), which estimates what a pitcher's ERA should be based solely on his strikeout rate, walk rate, and GB/FB ratio. These three components--K rate, BB rate, GB/FB--stabilize very quickly, and they have the strongest predictive relationship with a pitcher’s ERA going forward. What’s more, they are not very dependent on park effects, allowing us to make reasonable comparisons of pitchers across different teams.
The formula for QERA is as follows:
Note that everything ends up expressed in terms of percentages: strikeouts per opponent plate appearance, walks per opponent plate appearance, and groundballs as a percentage of all balls hit into play. Andy Pettitte, for example, has a 19.6% K rate, a 7.9% BB rate, and a 62.7% GB rate, giving him a QERA of 3.68. Note further that QERA is exponential, which is appropriate since run scoring is not linear. The QERA numbers used in this article are based on 2006 major league performance only.
We also need to have a way to account for the fact that the extra off-days in the postseason schedule allow a team to weight its outings toward its best starters. I looked at the distribution of playoff starts for all teams since 1995 that reached the World Series. It turns out that the average distribution is as follows:
#1 Starter 31% of playoff starts
#2 Starter 28% " " "
#3 Starter 23% " " "
#4 Starter 18% " " "
Of course, some teams go to greater or lesser extents to spot their best starters in a series, and their ability to do so will be partially determined by luck-of-the-draw factors like scheduling and rainouts. Most teams do not deviate too much from this pattern; the closest was the '98 Padres, which gave 13 of their 14 playoff starts to their top three pitchers. By weighting the individual QERAs by the percentages listed above, we can come up with an overall team score.
Projected playoff rotations are based on media reports within the past week. I’ve taken care to get these right, but I’m sure that I’ll get some of them wrong nevertheless, especially given how fluid things are in terms of which games will be meaningful in the remaining days of the regular season. These rotations are what it looks like the respective teams will do; that’s often quite different from what they should do. With all of that said, away we go:
1. Wang 4.58
2. Mussina 3.21
3. Johnson 3.81
4. Wright 5.17
Chien-Ming Wang is on 18 wins and counting, which when coupled with Johnson’s back problems, makes him a virtual lock to start Game One of the Yankees' LDS. Wang is one of those pitchers who, like Tom Glavine, continually manages to post an ERA that is far superior to his peripherals. Even after accounting for his superior groundball rate, the numbers say he’s a #3 starter, not a #1. What’s unusual is the way in which Wang is getting lucky. His BABIP is very normal, and he’s actually been a bit worse with runners in scoring position. So what gives? Wang has allowed just a .230/.271/.316 line to the hitter leading off the inning, and it's very hard to score runs when your leadoff man gets on only 27% of the time. There might be some element of skill in attacking leadoff hitters, and Wang is undoubtedly a smart pitcher who understands good situational baseball. Nevertheless, this has to be mostly luck, and the secret sauce reminds us that finesse pitchers tend to get creamed in the playoffs. This is a very vulnerable rotation, especially with both Johnson and Mussina nursing injuries.
1. Rogers 4.64
2. Verlander 4.34
3. Bonderman 3.16
4. Robertson 4.23
Some reports out of Detroit have Jeremy Bonderman dropped to the #4 slot, but I’ve promoted him to #3 after he made his third consecutive quality start against the Blue Jays last night, figuring that Jim Leyland will do the same. Indeed, a recurring theme for the Tigers is whether Leyland will deploy his personnel correctly, which in an ideal world would mean that Bonderman and Kenny Rogers would swap places and Joel Zumaya would inherit the closer’s job.
1. Santana 2.71
2. Silva 4.73
3. Bonser 3.55
4. Garza 4.89
The wild card here is Brad Radke, who will start tomorrow in an audition for the playoff rotation, potentially setting him up to displace Matt Garza. But this may be a trivial issue--Radke’s QERA is 4.43, which is virtually identical to his actual ERA (4.46), and it doesn’t make him a slam-dunk choice over Garza. Plus, there’s the risk that Radke is a late scratch or has to leave a game early, which can throw the pitching staff out of sorts for the rest of the series.
1. Zito 5.09
2. Loaiza 4.27
3. Harden 3.61
4. Haren 3.50
QERA really hates Barry Zito, whose K:BB ratio in the second half is an ugly 54:41. Indeed, Zito’s lack of command is doubly problematic, because it means he works a lot of deep counts, and that the 215 innings he’s thrown so far on the season are probably the equivalent of 230 or 240 for a more efficient pitcher. There’s an argument that if Barry Zito is your biggest problem, you have a pretty good rotation, but the numbers say that the starting pitching is not a comparative advantage for the A’s, even if Rich Harden stays healthy.
I'm putting Pedro Martinez in the #2 spot as a sort of hedge against his continued injury problems, but the real issue is in the back end of the rotation, where indications are that Willie Randolph will go with Steve Trachsel rather than John Maine. This is for no reason other than Trachsel’s cagey veteran-ness and the fact that he’s somehow managed to compile 15 wins on the season in spite of a 4.97 ERA behind a great defense in a pitcher’s park and awful peripherals. Remember the Maine! On the bright side, Orlando Hernandez has pitched very well since joining the Mets, and is no stranger to postseason success.
1. Hamels 3.32
2. Lieber 3.89
3. Myers 3.44
4. Moyer 4.67
Brett Myers is scheduled to start on Sunday, a game which will be a must-win for the Phillies under most scenarios in which they make the playoffs. I have listed him third here, but he might not be ready to go until Game Four of the LDS, and things could get further screwed up if Cole Hamels is needed to start a one-game playoff against the Dodgers or Padres. Nevertheless, this is a good and underrated rotation, and Hamels-Myers rates as the best 1-2 punch going unless the Astros can make the playoffs.
1. Carpenter 2.99
2. Suppan 4.72
3. Weaver 4.46
4. Reyes 4.47
We have a Johan Santana Lite situation here, in that the Cardinals pretty much need to find a way to get Chris Carpenter as many starts as possible. Carpenter pitched last night, and unfortunately for Tony La Russa, it looks like the Cardinals’ Sunday game might be meaningful, forcing Carpenter to take his turn and limiting him to one game in the LCS. Under that scenario, the Cards are a great candidate to get swept.
1. Oswalt 3.34
2. Pettitte 3.69
3. Clemens 3.27
4. Hirsh 5.69
There’s no discussion of the Astros' prospective playoff rotation anywhere on the Internet, for the obvious reason that there was no discussion of the Astros making the playoffs until approximately 36 hours ago. In any event, it will be the same shtick as last year, with the dominant front three and Jason Hirsh serving in the Brandon Backe Memorial fourth starter/enthusiastic bench guy role. Note that QERA is forgiving of Andy Pettitte, who has been widely perceived as having an off-year--you can bet that the Mets are pulling for the Cardinals.
1. Peavy 3.13
2. Young 4.35
3. Williams 4.72
4. Hensley 4.38
Speaking of forgiving, QERA doesn’t have any problem with Jake Peavy, who it rates as the third-best starting pitcher in the playoffs behind Santana and Carpenter. Chris Young’s numbers might be skewed here, because of his extreme flyball tendencies and his home ballpark. If the Padres lose the division to the Dodgers, and have to settle for the wild card, it might make sense to hold Young back until Game Three, when he’d be pitching at PETCO.
1. Lowe 3.77
2. Penny 3.81
3. Maddux 3.87
4. Billingsley 5.36
Chad Billingsley is a 1.2 IP, six-run disaster waiting to happen, but apart from that this group is solid. Not that the Dodgers have too many alternatives to Billingsley, unless they want to do something creative like a tandem start involving Jonathan Broxton. A three-man rotation is probably out of the question, with an aging Greg Maddux and fragile Brad Penny in the group.
One problem with a quick-and-dirty metric is that it's, well, quick and dirty. Although strikeout rate is not very dependent on park effects, it is dependent on league effects, and the National League teams should probably be docked 10-15 points because they don’t have to face DHs. Either way, there is no clearly dominant group of playoff starters this year, but the Astros, Phillies and Twins come closest.
Nate Silver is an author of Baseball Prospectus.
Click here to see Nate's other articles.
You can contact Nate by clicking here