A few weeks back, I prepared a post that compared Jaroslav Halak’s 2010 playoff series performances against those turned in by other Montreal Canadiens goalies, going back to Key Dryden’s rookie season. While I was happy with the piece I did, I had a lingering sense of doubt after it went out. Something was missing; the analysis was too simplistic, too cursory.
Later it came to me: in ranking Canadiens’ playoff goaltending performances solely on the basis of nominal save percentage, I was leaving at least three important factors out (there is a fourth factor, shot caliber, but there just simply isn’t enough data available prior to 2005-2006 to tackle this one). So, I went back and ranked each playoff series performance according to these three factors, and then created a weighted average of the three to arrive at the final performance metric. (Before I did this, I set the qualifying criterion at a minimum of 4 starts in a given series. That means that Steve Penney’s marvelous little 3-game run against Boston in 1984 is excluded from consideration.)
Here are the three factors, along with their respective weights:
1) Shot intensity (weight = 25%): Human fatigue plays a critical role during any 60-minute sporting event. A goalie that faces 18 shots and makes 17 saves gets credited with a save percentage of .944. A goalie that faces 38 shots and makes 36 saves get credited with roughly the same figure. But being able to maintain that save percentage over those 20 additional shots should count in that latter goalie’s favor. To put it in economic terms, the buildup of fatigue assigns an increasing marginal cost to each extra shot faced by the goalie, so—all else being equal—shots 34 through 38 should be tougher to stop than shots 7 through 11. Of course, there is the concept of the athlete who “gets stronger as the game wears on,” but I have to make the assumption that those athletes are special cases.
2) Opponent’s regular season goal differential (weight = 25%): On balance, the average shot coming off the blade of an elite sniper is more difficult to stop than the average shot coming off the blade of a 4th-line grinder. You can label that the God-given talent factor. Unfortunately, detailed shot-by-shot data doesn’t exist for most of the time period I covered. So, by way of proxy, I looked at the opposing team’s goals-above-league-average (expressed as a % of the league average) for that given NHL regular season. For example, in 2010, the Washington Capitals scored 318 regular season goals, while the NHL average was 233; so, their regular season goal differential (expressed as a percentage of the NHL average) would be 36%.
3) Series save percentage versus league average (weight = 50%): Call this my attempt to account for inflation. Improvements in goaltending equipment and goaltender technique, along with a series of other factors, have combined to push the average NHL save percentage up almost uninterruptedly since the league began tracking the metric officially back in the early 1980s. Put differently, posting a .910 save percentage back in 1984-1985 was quite a big deal, whereas it would have been fairly commonplace in 2009-2010. So, I calculated the delta between a given goalie’s series save percentage and the NHL average for the playoff year (the aggregated data comes courtesy of QuantHockey.com).
And here’s how the re-crunched data comes out:
In the end, Jaroslav Halak’s 2009-2010 performance against Washington comes out as the single best playoff series performance by a Montreal Canadiens goalie since 1983-1984. It’s a pity that I can’t push this back into the Dryden era, but I do feel better about this analysis than I did about the first iteration. While my method might not be perfect (having no way to measure shot quality across the years is quite unfortunate), I think you will find it to be far more rigorous than my first go round.