"On Corsi and Faceoffs" Rebuttal

Tom Benjamin over at Canucks Corner decided to take a swipe at the stats crowd yesterday, and usually I leave the rebuttals to Gabe but this one is relatively easy.

The first sentence of his blog post: "I'm fairly skeptical of much of the statistical analysis being kicked around these days."  So you'll have an idea where this is going.  Basically, Benjamin feels Corsi ignores team context and that is a big strike against it.  An appropriate conclusion, if you're Tom Benjamin and you ignore the fact that hockey statisticians have spent years developing a number of measures to adjust for and otherwise acknowledge team context.  Other useless critiques, such as "the sample size is always too small" and "explanations can't compensate for a number that ranks Ryane Clowe as one of the best players in the league" are pretty typical of what we've been hearing.  Typical as in assuming we're morons throwing around imaginary numbers with little thought.

Of all people, statisticians are aware of their sample size; a statistician that attempts to make definitive, over-arching statements with small samples is what you would call a bad statistician.  Corsi wasn't established as a valuable statistic with a small sample size; it's been vetted and adopted by some of the best hockey statisticians out there, including Gabe, Tom Awad (who used it to develop his Delta metric), JLikens, Vic Ferrari, Derek Zona, Jonathan Willis, Rob Vollman, and countless others for whom I've not the energy to find examples.  More than anything, what these non-moronic people have found is that Corsi needs to be given context, as who you face and who you play with can affect your Corsi outcome.  This is precisely why so much attention was paid to the Manny Malhotra/Ryan Kesler deployment last year; it was a great move, because both players could handle tough assignments, but the signing of Malhotra added a Corsi-proven centre to take the load off Kesler's shoulders.  When I'd say Malhotra is one of the best defensive forwards in the game, and somebody said, "But look, he only managed a +9 on a team with an excellent goal-differential", I could respond that he had tough assignments and I could provide solid evidence if the discussion needed to go further.  Now, if you're not concerned with backing up your statements, or if you're so full of yourself that you assume people will view your opinion as an expert opinion not to be questioned, then by all means make definitive statements based on your observation alone (which, incidentally, is a VERY SMALL SAMPLE SIZE).

For the rest of us, who might feel that a statement needs a bit more verification than "My card reads 'Hockey Expert,'" we both observe and seek to verify those observations with more-objective data.  We used Corsi because study after study came back telling us that shot-based metrics reveal the most about an individual player's skill offensively and defensively, and it helps us build the context that Tom Benjamin so desperately seeks.

To the second point about Ryane Clowe: it's a common misconception that, because Clowe had a great Corsi performance last year, a Corsi advocate would call him the "second best player."  Well, that's true if we were willing to ignore the Corsi-based context, which says that he faced average competition and played on a line with the best possible Sharks forwards, but we wouldn't do that.  Regardless, to call someone the "second best player" alludes to talent, whereas to call Clowe's 2010-11 as the "second-best player performance" would be more accurate.  We need a bit more of a sample from Clowe before we start calling him anything more.  The funny thing is that Benjamin seems to be fighting himself with this example, as he previously felt that Corsi does not provide enough information about a player, yet he can't see how a player such as Clowe could be so valuable based on the boxcar statistics that he likely relies upon (plus the observations he's made).  He wouldn't be able to understand why Gabe would shower Samuel Pahlsson with such praise (read the last paragraph), either.  Well, maybe there's somebody here who needs a bit more information.  

P.S. He likes Tyler Dellow's work on faceoffs, apparently, which he prefaces by saying that he's already reached a similar conclusion but according to him that's not why he likes the post.  Is that like when a white person says they have one African American friend?

P.P.S. Benjamin, like SI's Brian Cazeneuve before him (who was rebutted by another advocate of Corsi, Kent Wilson), believes very strongly in the word "puck possesion" "puck position."  If only we could find a metric to help define it...