By Matt Pfeffer
Player turned analyst Jeff O’Neill caused a stir this week with a jab at corsi, one of the new-fangled hockey statistics:
Colton Orr had the best Corsi on the ice in a game that had Kessel, JVR, Crosby, Malkin. Let’s stop the madness.
— Jeff O’Neill (@odognine2) October 27, 2013
Colton Orr, as it turns out, had himself a fine game. Orr was on the ice for nine of his own team’s shot attempts and only one (one!) of the Penguins’, for a 90 CF% (corsi for percentage). Nobody in the fancy stats community is suggesting Orr is a better hockey player than Phil Kessel, Evgeni Malkin or Sidney Crosby. It’s silly to think any stat will display a player’s true talent level in any one-game sample size. But let’s not forget something: Orr was on the ice for nine shots for and one against. That’s amazing. It’s not something he (or any other player) can replicate consistently, but that shouldn’t take away from his performance.
It should make intuitive sense to all hockey fans, geeks or not, that being (partly) responsible for such a large difference in his team’s shot differential (the Leafs were out corsi’d that night by a score of 39 to 28), is undoubtedly a positive. But I think performances like this should be viewed with the same good-heartedness of an unlikely goal scorer producing for his team, not as evidence the statistic is flawed. Corsi though, unfortunately for O’Neill’s argument, does in fact correlate rather well with what we perceive as quality hockey players in larger sample sizes, just like goals and points.
I think there is a divide between the online hockey stats community and those who prefer more traditional stats. There doesn’t have to be. The concept of shot differential, quality of competition, zone starts and WOWY (without you and with you) analysis can only further help our interpretation of things like goals, assists and points. These ideas have always been floating in the back of our heads when we’ve looked at hockey stats before and now we can quantify them.
Many think advanced stats are a conversation-ender. If someone says Player X is good and you point out the player has a bad corsi, you are contradicting him and saying he’s wrong. This isn’t or shouldn’t be the case. The kind of conversations we have about offensively unproductive players and what value they bring is the exact same that should be had for corsi. If a player is allowing an unfavourable differential of shots while on the ice, why is this happening? And why is this not as bad as it seems? What is shot differential not telling us about this player? These are the kinds of conversations that will help push our understanding of hockey further.
With nothing more than a computer, any hockey fan can go on to any number of sites and find a ton of information on players and teams they’ve likely ever seen before. The expansion of hockey stats over recent years has been amazing – and it’s accelerating all the time. With a ton of new stats comes a ton of new ways to think about hockey and hockey players in ways you’ve never thought before.
When I watched hockey growing up I never would think to consider the impact of how a player is used and how that might affect his play. Now, being exposed to QoC and Zone Start metrics, I can’t help but think about a game in the context of who’s out on the ice against each other and when. Advanced stats will change and evolve the way you think about hockey, and you should think about them like you would with any stat: take them with a grain of salt, not buckets of it.
There are many great websites where you can find advanced statistics for the NHL. HockeyAnalysis.com, ExtraSkater.com and BehindTheNet.ca all do excellent work. I will be launching my own stats site in the coming weeks called PuckCharts.