Please visit Puckerings for more hockey stuff by me   

There are certain goaltenders who have reputations as “money goalies”; that is, they perform better when it “really counts”. Their regular-season stats are good, but they really “turn it on” for the playoffs. Goalies with “money” reputations include, for example, Tom Barrasso, Ed Belfour, and Bill Ranford.

On a superficial level, there seems to be evidence that this is true. Here are the career save percentages for these goalies in the regular season and in the playoffs, through 1999/2000:

All three apparently perform substantially better in the playoffs. But wait; here are the same figures for all goalies who have played 30 or more playoff games through 1999/2000, and for whom career save percentage is fully available:

We have a sample of 20 goalies. Only two have lower playoff save percentages than regular-season save percentages. On average, theses goaltenders save percentages in the playoffs are 1% higher than in the regular season. Therefore, we cannot call Bill Ranford a “money goalie” due to a 1% increase in playoff save percentage, because it seems save percentages in general increase 1% in the playoffs.

Looking at the goalies with the biggest improvements, Kolzig and McLean would probably be considered “money” by some. But Ken Wregget? Does this mean he is unfairly not considered a money goalie? Or does this mean that these numbers are quite flukey, due to the small sample sizes? I’ll take door number two, I think.

This underlines the importance of context. Things are often said without a hint of context. “Toronto is 18-0-3 this year when leading going into the third period.” So what? Is this better than most teams? How big a lead do they usually have? Do you expect me to accept an inference made on a sample of 21 games?

Analysis is meaningless without context. Bill Ranford is not a money goalie just because his save percentage is 1% higher in the playoffs. When a claim is made, you need to identify the context of the numbers used before you can assess its validity. I guess that’s my point.