Howard Smith-US PRESSWIRE
In 2012, Cliff Lee and Cole Hamels had very similar seasons. Statistically, that is.
I’d like to think that most people would see Cliff Lee’s and Cole Hamels’ 2012 seasons as a kind of death knell for the pitching “win” as a thing. One (it doesn’t even matter which) threw 215 ⅓ innings, another 211; one gave up 79 runs, one 80. One posted a 131 ERA+, one 127. In runs-allowed terms, it’s hard to find two more similar pitchers; by the various advanced stats that try to remove luck and defense (FIP, xFIP, etc.), Lee came out just a bit better.
Two left-handed pitchers have the same performance for the same team, and Hamels goes 17-6 while the at-least-equal Lee goes 6-9. It seems clear that these divergent records can’t tell us anything. One reader of Bill James’ site (laughed off by James -- a friend of his, he explains in response to a later e-mail -- but picked up and tossed to his readers by Tom Tango) disagreed. The gist of his argument was that Lee and Hamels were pitching in different run contexts. They pitched for the same team two or three days apart from one another, yet the reader was convinced that Cliff Lee’s numbers look better than they were because he pitched in games that, for some reason, it was harder to score runs than on days when Hamels pitched. His argument was that won-lost record was, “the only stat automatically adjusted to game conditions and environment, etc,” and thus “had some value,” because it reflected the fact that Hamels pitched better (and in tougher conditions) than Lee.
There are a number of reasons that this seems kind of silly. The most obvious is that Hamels’ winning percentage was nearly twice Lee’s, which to me means the won-lost records suggest Hamels was about twice the pitcher. The reader acknowledged that Lee was a good pitcher who had been unlucky, but in order for the records to mean anything, doesn’t the gap between the two have to be reasonably close to the W-L gap? And that just doesn’t pass the straight-face test.
Nonetheless, I thought it’d be interesting to take a stab at divining whether there was any overarching difference in the “run contexts” Hamels and Lee pitched in. I can think of several things that might lead to two pitchers on one team pitching in different run contexts, and I created a spreadsheet -- which you can see here -- showing the conditions the two pitchers faced on a game-by-game basis, or at least those we can measure; I can’t tell you how they might have been affected by their choice of pre-start lunch:
Temperature and precipitation: conventional wisdom, which we’ll accept for these purposes, is that warmer temperatures and calm weather are good for hitters, and that cold, nasty weather is good for pitchers. Neither Hamels nor Lee pitched in a game this year in which Baseball-Reference’s box score noted any precipitation, so temperature is the only consideration here.
Wind speed and direction: I assume that high wind speeds in any direction are a little bit of a negative for a pitcher -- making it harder for the defense to catch the ball -- and it’s pretty self-evident that it’s a lot worse if the wind is blowing out.
Park factors: You might figure a team’s no. 2 and no. 4 pitchers would see a lot of the same teams and the same parks, but both Hamels and Lee pitched in several parks the other did not. If Lee appeared in more pitchers’ parks than you’d expect, he could fail to win a lot of low-run games, but not actually pitch that well, and it wouldn’t necessarily be reflected in his stats (save the W-L record).
That’s all I can think of that we can measure. The strength of the opposing team and lineup shouldn’t matter, because the idea is that the Lee conditions were such that neither the opponents nor the Phillies could hit worth a damn, leading to all those losses and no-decisions. There’s one other possibility (mentioned in the comments to Tango’s post) that I can’t measure: if Lee happened to pitch to a ton of pitcher-friendly umpires, and/or Hamels to hitter-friendly ones, that could certainly make Lee look artificially comparable. Lee pitched for 26 different umpires and Hamels 25, so if you want to dig through that mess, you have my blessing and my pity.
Looking at what we do have, here’s what I can come up with:
Lee may have pitched in very slightly warmer conditions overall, with an average temperature 0.3 degrees and a median two degrees higher than in Hamels’ starts. Lee’s first three starts of the year were each colder than any one of Hamels’ starts, but he also had five 90-degree starts to Hamels’ two. This is a wash at best, and might actually be a slight advantage to Lee for having to pitch in more oppressively hot conditions.
The average and median wind speeds for their starts are almost exactly identical; nothing to learn there. However, the wind was blowing out in 17 of Lee’s 30 starts, compared to only nine for Hamels. Here again, if anything, it’s slightly better news for Lee.
On park factors: both had fourteen starts (and between 96-100 innings) at home, so CBP’s 101 park factor (using the multi-year “pitching” park factors available on Baseball-Reference) would add only noise; I removed those and looked at the rest. The median road park factor for both pitchers was 100, but the averages come out to about 101 for Hamels, about 100 for Lee. So Hamels may have pitched in slightly tougher parks. I don’t think we’re close to making Hamels twice as good yet.
And that’s it, right? I also looked at the sky conditions, and it’s actually kind of odd; Hamels pitched in a dome five times, and of his 26 other starts, Baseball-Reference classifies 25 as either “Cloudy” or “Overcast,” compared to 21 of Lee’s 28 outdoor starts. Now, Most of Lee’s “clear” starts (and Hamels’ one) were at night, so neither got much benefit from the glare of the sun, but you could give Hamels a bit of credit based on that if you wanted to.
Finally, the Tango comments gave me another idea: if it’s the game conditions helping Lee out, then the bullpens would likely benefit from the same conditions, so the spreadsheet tracks that, too (very simply, using only ERA), and as you might expect of one who pitched brilliantly and won six games, Lee was backed up by an awful bullpen, posting an ERA very near 5.00 in 60 ⅔ innings; the same bullpen was brilliant for Hamels, with a 2.39 in 64 innings. So if there was some invisible set of factors such that Lee was pitching in more favorable conditions than Hamels was, it appears that those conditions immediately reversed themselves once the starters were pulled.
This is a long-winded way to argue something that really should be obvious by now, and probably is obvious to most readers here: that wins and losses are impacted by too many other factors for a pitcher’s single-season won-lost record to say anything meaningful about a pitcher’s performance. There’s a lot we don’t understand about pitching, and I think there’s probably a lot we don’t understand about Lee’s bizarre season, but the wins and losses just don’t get us any closer to figuring it out.