Radical Stat Proposal: The Own-GoalFebruary 7, 2012, by
I understand I may be a little biased when it comes to this subject, being a goalie and all, but before you judge, hear me out. Think about how many times during the course of your hockey watching career you have seen a goal hit an unsuspecting defender’s skate/pant/shin pad and go in behind a goalie who should have been innocently watching that puck sail wide. Should the goalie be responsible for this? If the goalie does his part to keep the shooter from putting the puck into the net only to be foiled by his own teammate, should the stats reflect his “failure”?
Therein lies the crux of the issue. If you were to quantify the number of times the puck inadvertently bounced off the defenseman into the net in his own stats, as an evaluator, can tell that number most likely represents a fluky sample. However, it is much more difficult to extract those types of situations from traditional goalie stats. Let’s say hypothetically that our sample defensive group is guilty of 12 own-goals for the season. Unless some deeper statistical tendencies (a subject for another post) could be derived from the number, you can see that 12 instances of bad luck found their way into the back of the team’s net. If those twelve goals were embedded into the rate stats for the goalie for the season, it changes the equation significantly. For example, last season Hank played 66.8 games (4007 min.) and had a very respectable 2.28 GAA. He allowed 152 goals during that time. If you subtract our own-goal sample out of 12 goals, it brings his goals allowed down to 140. His GAA goes from 2.28 to 2.10. His save % goes from .923 to .928.
This exercise works for two purposes. First, is proper allocation of responsibility for what occurs on the ice. While it may seem harsh to punish the defensemen tying a guy up in front of the net who falls victim to a bad bounce he probably couldn’t have avoided, this happens all the time with the +/- statistic. Back-checking forwards can be victimized by the dreaded minus because a 2-on-1 goal is scored. Seems unfair, right? The goalie playing the position properly should not be held responsible when a shot, which is not a shot on goal, falls into the back of his net due to one of his teammates.
The second purpose is one of evaluation. As I mentioned before, it is much more difficult to dig these fluky situations out of goalie’s rate stats. All they do is serve to skew the perception of the goalie’s abilities when examining these stats. Also, if one defensemen has an excessive amount of own-goals over the course of a large enough sample, it may tell you something about that defensemen’s positioning tendencies. Assuming you can draw information outside of small sample-size noise, it could be a useful tool in improving defensive positioning.
The NHL will probably never adopt the own-goal statistic, but I think that it makes a ton of sense. From a defensemen’s standpoint they are an isolated incident. As long as they fall within an average statistical range, they can be dismissed as bad luck. For the goalie, it can be the difference between a good season and a Vezina season. As the statistical evaluation models become more advanced and we start to quantify more minutia that happens on the ice, I still hold out some hope we could employ this type of stat to understand that line that blurs skill and luck.
* For clarification purposes, the scoring on the own-goal would still be a goal for the last offensive player to touch the puck, but would be treated like an empty net goal for the purposes of goalie stats and credited to the defensive player whom the puck deflected off of.