Hockey News Makes a Mess Out of Stats

In case you haven't heard the Hockey News has been tracking their own version of "clutch scoring" this season. If this were evidence that the paper is embracing advanced hockey stats I would be thrilled. But the truth is all they have done is slap something together that any hockey loving middle school student could create.

A new wrinkle on Campbellnomics this season is the comeback goal. A comeback goal can only be scored when a team is trailing by two or more goals and that goal has a direct effect on his team getting back into the game. The goal must be one of goals scored in succession that result in the game later being tied.

So if the Thrashers are down by two goals and Bryan Little making an amazing play and scores to cut the lead down to just one goal--but the Thrashers fail to tie the game--that goal is not a "clutch goal" because the other 17 skaters failed to tie the game up. Does that make any sense? On other other hand if Boris Valabik throws a puck at the net and it deflects off someone's stick and then redirects off another players butt and into the net--and cuts the opposition lead to one and the Thrashers go on to tie the game--well that counts as a "clutch goal"---greeeeat! That's just sheer idiocy.

A player can't possibly know how a game will turn out--all they can do is play their best given the situation. If they score a goal that helps their team no matter what the score is. To assign special emphasis to goals based on how the game turned out makes little sense.

If you want to make a case that scoring the sixth goal in a 6-1 is less meaningful than I would agree with that. Some goals are more "high leverage" than other goals. You could use a "marginal goals framework" in which you look at the historical average and given credit to players who score in high leverage situations.

For example, in a tie game the next goal will always be meaningful and the closer that tie game is to the end of regulation, the more meaningful that goal will be. A goal that gives a team a 4 goal lead is usually less meaningful, but there is a chance the opposition will score 3 more and the goal that produced a 4 goal lead will suddenly become the gain winner. That goal that gives your team a big lead is unlikely to end up as the game winner, but you never know.

If the Hockey News wants to do "clutch scoring" correctly they ought to assign a value for every goal scored based on the game score and the time at which that goal is scored. Some goals have a HUGE effect on the probability that your team will win the game and other goals not so much. The Hockey News ought to ask one of the quantitative bloggers out there to help them produce a more accurate measure of "high leverage" goal scoring events than mish-mash approach they are taking with their amateurish attempt.