The hirings of Kyle Dubas as assistant GM in Toronto and Tyler Dellow as analytics expert in Edmonton have sent hockey’s good ol’ boys network into a panic. The contingent of hockey pundits who tout “intangibles” have long suspected and feared a reckoning at the hands of the spreadsheets. Now that it has finally happened, they effetely plead that the new era of analysis would not begin and end with Corsi:
Here's hoping Oilers don't go all-in on flawed Corsi analysis. But I expect more from Dellow than that…
— David Staples (@dstaples) August 5, 2014
These people miss the point. Advanced statistics aren’t just one piece in a bigger conversation. Advanced statistics are the end of the conversation.
Named for author Jerome Corsi, Corsi counts the number of corsis that players are on the ice for during hockey games. That number, called a C.O.R.S.I. number, is the best way to evaluate player skill. More corsis, better players*.
It has been proven, over and over again, that things like heart and clutch play and even random chance cannot hold a candle to digital incandescence of Corsi.
Granted, this was not always the case. Corsis used to be tracked by hand. And back before we knew of score effects, zone starts, and THoR, Corsi and its brother stat (simply called MATT) were certainly inferior to the Eye Test. But now that the methodologies and technologies have matured, Corsi is almost** perfect, as proven by the recent glut of geek hirings. The ever-shrinking “watch the games” crowd is now left to do inconsequential things like watch the games.
Dellow, Dubas, and even Eric Tulsky (who has also been hired by some team, no idea which, yeah) first made their reputations by exposing the eye test as irrelevant to the task of hockey evaluation. This was unwelcome news to coaches who ran video sessions and hockey writers who didn’t know the difference between a TR and a TD, but it was the truth (with a .553 Pearson correlation coefficient!). The automatic tabulation of shot attempts and zone entries by computers is simply more reliable than human vision, which can be hindered by astigmatism and diabetes mellitus. Still, the adjective-driven hockey media clings to its outdated way of life like a triceratops munching on a fern in the Yucatan at the end of the Cretaceous period.
That’s not to say the soon-to-be-extinct members of the numbers-free press are useless. They’re not; they are merely lone wolves… for now. But once the SportsVU system gets implemented in the NHL, the opponents of Big Corsi will be rendered utterly meaningless.
Some think that’s a bad thing. They cling to their worn-in, dog-earred copies of The Game and Open Net with fearful nostalgia. The game, they say, will one day be reduced to nothing but cold, hard numbers.
Those people are wrong. We are already there.
The dirty little secret of the stats community is that hockey can already be expressed entirely by statistics. The power elite of CoRSi’s inner circle actually have the power to simulate entire games using nothing but mathematical models. While most of us are suffering through the tedium of a hockeyless offseason, the fancystats oligarchy can generate and enjoy hundreds of hockey seasons using just a laptop and Microsoft Excel.
Hockey stats were first invented to make the sport comprehensible. At the time they were just one tool of many, intended to be used in concert with qualitative analysis and the whimsy of opinion. But as technology improved, those stats– especially C0RSI– began to eclipse the traditional styles of talking about hockey. Now that the purveyors of advanced statistics are finally ensconced in hockey’s power elite, the conversation can finally end.
* Except in the case of defensive defensemen, if convenient.
** Nevermind; it actually is perfect.
Russian Machine Never Breaks is not associated with the Washington Capitals; Monumental Sports, the NHL, or its properties. Not even a little bit.
All original content on russianmachineneverbreaks.com is licensed under Creative Commons Attribution-NonCommercial-ShareAlike 3.0 Unported (CC BY-NC-SA 3.0)– unless otherwise stated or superseded by another license. You are free to share, copy, and remix this content so long as it is attributed, done for noncommercial purposes, and done so under a license similar to this one.