Monday, June 06, 2011
Pitch ‘em when and where they’ll winPosted by Will Hatheway at 4:04am
Baseball exists in a sea of anecdotes, some of which ultimately to transform into anecdotal evidence that may or may not point to the truth. That is what makes sabermetric analysis so great, in my mind: It is the scalpel by which we can dissect conventional wisdom and see whether it is based on solid ground. Many aspects of the game, including those that are applicable to the fantasy baseball enthusiast, have been covered, but I found myself thinking about one that I’m not so sure has gotten the full treatment: streaming pitchers.
Streaming reminds me of trying to time the stock market, and, as with that financial correlative, I initially felt uncomfortable with the idea. I am, after all, a “value” guy … in fantasy, I love auctions because you can fill the “scrub” portion of your squad with single-digit gems, while in finance I fawn over companies with low price to sales and free cash flow (that are to price-to- earnings as FIP is to ERA).
The problem with the stock market theory, however, is that I have often found my “brilliance” unappreciated for quite some time, leaving my money to either tread water or even sink down. Why? Because it just so happens that, at least in the near term, what’s going up tends to keep going up, while what is falling tends to continue to fall. In other words, momentum exists, so I finally wised up and accepted that “value” isn’t all that valuable until you pair it with a sort of market timing.
Similarly, seeking fantasy value in starting pitching sensibly begins with looking at the better metrics (FIP, etc.) as a way to see what those looking at the surface don’t. But we can do better, by which I mean that we can play the broader market swings, as it were, in choosing when to employ our bargains on the active roster. This is hardly revolutionary, but I have never seen actual numbers to back up streaming strategies. Since I am currently in a position in one league where I really want to employ it, I figured I ought to poke around and see whether it really provided a value-added product.
Here’s the deal: I am mired mid-pack in a saber-oriented points league despite having a strong pitching staff. The thing is, not only do I have a trio of studs (Clayton Kershaw, Cole Hamels, and CC Sabathia), but I also have a cornucopia of others from which to pick and choose when to start whom. (Eight others, to be exact, which has already brought out Brad Johnson’s scorn in a comment thread on this site, but that’s neither here nor there!)
We only have only nine active offensive spots, so I’m not too concerned with having a deep bench of bats, but I do, of course, have to take care not to waste innings, since we are capped at 1,400 (I am a bit over max-pace). What I wanted to know, then, was whether I could take the guessing and otherwise emotional/subjective bases for choosing who starts when, and instead rely on a simple but rational system for doing so.
In short, I think I have. I came up with a quick and dirty way of choosing exactly what makes for a very favorable match-up and then figured out how well pitchers did in each of their games in either those favorable situations or in those that didn’t qualify as such.
First, the criteria (I said they were quick and dirty, and they are, but it’s something): I used Cairo’s final preseason projections for team runs scored so as to exclude the above average offenses. I then turned to Stat Corner to exclude those parks whose wOBAs were unfavorable to my pitchers (this included taking into account handedness, such that righties wouldn’t start, for example, in Cleveland or Citi Field, but lefties would). I then took each pile and came up with a collective WPA (for saber leagues) and ERA (for 5x5s). The result? The predictably pitcher-friendly match-ups deliver, on average, a positive .115 WPA and a 2.44 ERA, while the other group had a NEGATIVE .024 WPA and a 4.40 ERA.
Now, I know this isn’t rocket science, but I’ve just never seen it tested using actual numbers. There is, of course, one clear problem: Combined with always starting my relievers (since they basically have near-ideal pitching situations every time out), just starting my guys (all 11 of them) when the stars aligned just right would leave me barely halfway toward a maximum innings-pitched pace. So what I did was add back in to the mix the sub-optimal starts from my three aces, putting me right around the innings cap. The result? A still lovely 2.88 ERA across 388 innings (412 would be on pace to pitch the maximum, so it still isn’t perfect, but close enough for me right now).
Now, this takes some patience (I would have started Jordan Zimmermann, Daniel Hudson, Matt Latos* and Bud Norris only once each since their home parks don’t favor them, while sitting the nicely performing Matt Garza the whole season to date… though those four combined starts resulted in ERAs of 0.00, 3.38, 3.00, and 4.50 respectively). And it takes discipline (I would have had to start Ted Lilly six times, more than any of my non-aces, despite the fact that he is basically sucking right now… though at least the resultant ERA would be .65 lower than the starts he wouldn’t be active for). But you can’t argue with the results, right? Repeat after me: a nearly full complement of innings at the low, low rate of a 2.88 ERA.
*What’s so weird is that Latos not only pitches in San Diego, but also in the pitching-friendly NL West, yet only one of his seven starts is considered “safe,”; perhaps that partially explains his 4.64 ERA in those six games?