Wednesday, December 13, 2017

Flaw in NHL Shot Suppression Statisitc

The modern age of NHL hockey has seen a rapid evolution of advanced statistics. We are entering a Golden Age of analytics. There are still flaws in some of the conclusions we are making from these metrics. One skill that current analytics does not properly value are defensive specialists. Players who specialize in preventing the other team's best players from scoring without providing much offense themselves will typically do very poorly in the "shot differential" battle, while still providing a useful service.

The term "shot suppression" in NHL statistics generally refers to "Corsi Against", or how many shots the other team averages per 60 minutes while a player is on the ice. If the opposing team averages more shots on goal while a specific player is on the ice, he is said to be bad at shot suppression. That appeals to our common sense. However, it is flawed in the context of line match-ups. If each line on a team averages the same amount of ice time against each line on the other team, then this suppression statistic is more valid. At the NHL level we see some notoriously complex line matching, where coaches attempt to deploy specific players against specific players.

If Player A gives up 40 shots per 60 minutes and Player B gives up 20 shots per 60m, we could say that Player B is twice as effective at suppressing shots. This might be correct, if A and B spend the same amount of time on the ice against the opponent’s top line(s). However, if A spent most of his time on the ice against Sidney Crosby and B spent most of his time on the ice against Riley Sheahan, then this "shot suppression" statistic is completely misleading.

Shot suppression should not be how well you do relative to your teammates. It should be how well you do relative to your opponent. Let's say the Crosby line averages 50 shots per 60 minutes and the Sheahan line averages 10. Then Player A allowed 10 fewer shots than his opponent averages and Player B allowed 10 more. In that context, we could say that Player A is twice as good at shot suppression. 

Does the opponent you are matched up against average more or less shots when you are on the ice? That's the true "Shot Suppression" statistic that I'd like to know. I'm not aware of anyone out there currently calculating this statistic.

No comments:

Post a Comment