Wednesday, February 17, 2010

Dave Cameron said what?

Earlier today, AK hipped me to a piece by Dave Cameron of FanGraphs on ESPN. It's Insider Only, sorry. Cameron is an author for whom I have a lot of respect and admiration. Unfortunately, I found Cameron's hypothesis, supporting proof and conclusion to be unsubstantiated, and I'd like to discuss it here.

Cameron opens his piece by regaling us with the greatness of Clayton Kershaw: a K/9 over 9, a power fastball and deadly breaking ball (Public Enemy Number 1, according to Vin Scully), and a 2009 ERA that was better than Cliff Lee and Johan Santana, and Kershaw isn't even 22 yet. Seemingly because of this fact, his age, Cameron writes:

Unfortunately for Kershaw and Dodgers fans, history suggests that this may be as good as it will ever get for the young lefty. In fact, given the success he has had in the majors at such a young age, he may have already peaked.

So now we have our hypothesis: history suggests that Kershaw may have already peaked, given what we know about other young pitchers. It seems like a reasonable claim, or at least one that should be easy to support or disprove with evidence. Unfortunately, I don't see that from Cameron in his two supporting claims.

1. Pitchers "defy conventional growth curves", often peaking quickly and never recovering the effectiveness of their early age. In support of this claim, he offers this evidence:

A. Flameouts: Mark Prior, Kerry Wood and Rich Harden: Prior flamed out, as did Wood. Wood returned and has found a role as a closer. Harden has been inconsistent, but is not a flameout.

B. Pitchers who lose velocity: Oliver Perez, Scott Kazmir, Felix Hernandez and Tim Lincecum.

Let's ignore the fact that Felix and Lincecum were probably in the top 3 of all MLB SPs last year, despite a loss in velocity, and move to his next supporting claim.

2. "In the past 30 years, 11 pitchers have rang up at least 180 strikeouts in a single season when they were 22 or younger" (sic). This is an interesting supporting claim, firmly rooted in evidence, although not necessarily proof of anything. 11 pitchers is, by definition, a small sample size. Sure, there isn't a decent statistical sample size in existence, given that only a handful of pitchers have accomplished what Kershaw has accomplished at his age, but that doesn't necessarily tie Kershaw to those other names.

With a sample size of 11 names, the reason for each flameout could be entirely idiomatic and particular to each individual pitcher. For instance, Wood and Prior are widely regarded to have been overworked. By all accounts, this has not happened with Kershaw. This doesn't prove anything, one way or the other. It does show, though, that when you're dealing with tiny sample sizes anything can come up and skew your results. Dwight Gooden is another name on that list. He had a cocaine problem. Is he necessarily the best comp for Kershaw? No.

Furthermore, for some reason Cameron has decided to restrict his "search" to 1980-2009. Why would he do that? I don't know. I do know that when you open it up to 1901-2009 you get very different results.

Bob Feller, Frank Tanana, Christy Mathewson, Bert Blyleven and Don Drysdale all meet the under 22, over 180K requirement. I think it's safe to say that they didn't all peak at 22, and went on to have successful major league careers.

Now, my counter-argument doesn't prove anything. We still have a sample size problem. But it does show you that Cameron's piece is woefully insufficient and doesn't come close to painting the real picture. Is it interesting? Yes. Is it something to keep in mind if you're a Dodgers fan? Yes. Is it proof? No. The first supporting claim is abundantly weak: he throws out 7 names and expects it to speak for itself. The second supporting claim is slightly more compelling, but fails to have the necessarily statistical relevance to support Cameron's conclusion. When you widen the search, removing the arbitrary boundaries of 1980-2009, you get a slightly different picture. As such, this conclusion is entirely unwarranted:

But the reality of history shows that he's more likely to get worse than to get better, and fans counting on Kershaw to win a Cy Young or two are likely to be disappointed.

This claim, that Kershaw is "more likely" to get worse than better, just isn't supported by the evidence. I like Cameron, and I respect his work, and that's why I'm left scratching my head.

No comments:

Post a Comment