Game Winning Goals


Driving home yesterday, I was listening to Hockey Buzz columnist Andy Strickland commenting on forward Brad Boyes of the St. Louis Blues.  Strickland’s a fairly good columnist, he breaks stories sometimes and of all Eklund’s writers he’s probably the one most worth listening to, but he said something that struck me as ridiculous: that Boyes wasn’t scoring clutch goals any more, and that it was visible from his drop in game-winning goals, from 10 to two.

When I got home I decided to see if that was the case, by seeing if Boyes’ game-winners had dropped off more than his goal-scoring,  I did that by figuring out what percentage of Boyes’ goals were game-winners since the lockout.  All seasons are pro-rated over 82 games.

Season Goals % Game Winners
2005-06 26 11.5%
2006-07 17 11.8%
2007-08 43 20.9%
2008-09 33 33.3%
2009-10 16 20.0%


The interesting thing to note about that list is how close the percentage from this season matches the percentage from Boyes’ breakout 43-goal year; it’s virtually unchanged. 

But that got me thinking about game-winning goals in general.  Strickland went on about how he doesn’t want the guy on his team who scores the sixth goal in a 6-1 win (seems to me that 6-1 wins are plenty rare that guys aren’t padding their stats with those sorts of goals, but I digress) but I’m not at all convinced that game-winning goals do a good job of showing clutch scoring.  For instance, consider a game between Team X and Team Y; Team X takes a commanding three goal lead but Team Y comes back and scores two in the third, losing 3-2.  The game-winning goal would go to the player who scored the third goal for Team X, despite the fact that four of those goals were clutch and the only guy getting awarded the game-winner was the one who didn’t score a clutch goal.

In any case, I decided to put my theory about game-winning goals to a small test.  I took all of the players who scored 20 or more goals in both 2007-08 and 2008-09 as my sample set.  There was a very weak correlation between players that scored a high percentage of game winners relative to total goal-scoring year over year (.103) but almost to the point of irrelevance. 

There’s a lot of noise there, and I’d guess that the correlation has more to do with players on good teams scoring more game-winners than with anything else.  I’m actually a believer in clutch scoring; I just don’t think game-winning goals express clutch scoring very well.

Comments (3)

  1. Ken Campbell at The Hockey News has done a “clutch scoring” thing for a while on their website. It has a stupid name, “Campbellnomics” or something similar, but to give credit where it’s due it actually factors in things like first goal of the game, goals in comebacks, goals that tied the score, etc., at the same rate as game winners, as far as I remember.

    It’s not perfect, but it’s not a bad measurement for clutch goal scorers either. The top guys are almost always at the top, I think, but even twenty players deep you can get some interesting names in the mix.

  2. Brian P: I’ve seen it, but I’ve never been sold on it. I’m a fan of statistics but I don’t think they do a good job of showing which guys are good in pressure situations because statistics need fairly large samples to give an accurate reading. The difference between playoff and regular season scoring is probably the best comparison, with proper adjustments (for opposition quality and the lowered goal scoring between the two) but even then I’d limit it to guys with big samples of time.

  3. This is true, guys who can maintain of even increase their scoring pace in the playoffs would probably be a good indicator over time for clutch scorers.

    I’ve heard tell of ex-NHL GM Mike Smith’s current company, who’s name I can’t remember, and Smith claims that he can tell teams who the clutch scorers are with some kind of large scale statistics analysis. I don’t know if it ever took off, but it would be interesting to see the kind of data it came up with someday.

Leave a Reply

Your email address will not be published. Required fields are marked *