Optimizing the Nationals Rotation
We've tried (somewhat inconclusively) over the past few weeks to find out the optimal way to organize the outfield and to arrange the line-up. Now is the time to take a crack at pitching.
Similar to the outfield situation (where we must consider positional adjustments, defense, and offense) and the batting order (where the trade-off between having more men on base and getting better hitters more PAs remains mysterious), optimizing the rotation is not straightforward, though more so than the former tasks. There are two things that will improve a pitcher's WAR: having a low FIP and throwing a lot of innings per start. While FIP dominates WAR, eating a lot of innings can also contribute; John Lannan's poor FIP projection for next year (4.58) is mitigated by his high innings per start rate (almost 7).
Even with his inning-eating ways, Lannan still does not make the optimal rotation. That rotation consists of Ross Detwiler, Jordan Zimmermann, Tom Gorzelanny, Craig Stammen, and Garret Mock; this is the same rotation you would get through sorting by FIP. Using Bill James projections, innings per start from the last season, my shaky understanding of WAR, and a bit of optimism concerning how many starts these guys will make, I am sort of confident in saying the optimal rotation would be worth about 10 WAR (with more wins potentially coming from spot-starters and whoever else starts games next year). Compare that to the net WAR of 8.5 from last year's starters. Or the predicted 8.68 WAR from what I think the actual rotation will be (Livan Hernandez, Zimmermann, Gorzelanny, Jason Marquis, and Detwiler).
Not content with these mundane rotations, however, I decided to test a few others. First I did the Gottfried Leibniz rotation a.k.a. the Best of All Possible Worlds rotation. In this rotation Stephen Strasburg repeats his dominance of last season but sustains it for 30 games (6.81 WAR right there), Yunesky Maya pitches decently (4.00 FIP), and Detwiler, Zimmermann, and Gorzelanny do as expected. This comes to a total of 15.41 WAR. Dr. Pangloss, I presume?
Then, of course, there is the worst possible rotation: Luis Atilano, Marquis, Lannan, Hernandez, and Mock—for a grand total of about 5 WAR. Less hyperbolic, but still interesting, is a rotation in which the Nationals actually complete that deal for Fausto Carmona: this would not change the optimal rotation but would add about 1 WAR to the “actual” rotation.
Now, I know that none of these things, possibly not even the “actual” rotation, will come to fruition, but even if the Nationals did try to optimize WAR (something that should never involve making Livan Hernandez the ace) they would still fall far short of what could possibly be the most WAR-optimizing rotation ever: The 2011 Phillies rotation, which I project will go for about 20 WAR (which could actually be conservative). Even with Strasburg at the top of his game, the Nationals could only put up 15 WAR, and that is very optimistic.
This exercise, of course, like those found in the linked articles, is academic. I know the Nationals will start Livan on Opening Day because that's what they're supposed to do—I mean, I guess he did have kind of a miraculous year last year. But the Nats should think about taking risks (I guess these risks would include “looking weird” or getting questioned by the scary mainstream press) to procure a few more wins. After all, for about thirty years Bill James seemed to outsiders as though he was perfectly insane. And then the Red Sox won the World Series twice. Let's hope the Nationals front office was paying attention.
Note: As always, if you want my data, please e-mail me ( This email address is being protected from spambots. You need JavaScript enabled to view it. ). My Excel worksheet will also let you calculate your own WAR values!
