Please visit Puckerings for more hockey stuff by me  

The 1972 Canada USSR Summit Series is often described as the greatest hockey series ever played. Certainly, the goaltenders involved could hardly have been better. Vladislav Tretiak, Tony Esposito and Ken Dryden are certainly three of the greatest goalies of all time. Of course, Esposito and (especially) Dryden did not play well in the series, which kind of cheapened the experience.

How bad were the Canadian goalies? It’s hard to say, exactly, because shot data is available only for the first five games of the eight-game series. So I thought it would be fun to make an educated guess at the shots in the final three games, using the data from the first five, to arrive at an estimate of each goaler’s performance.

Here’s what we know:

So what we want to do is estimate the shots in games 6, 7, and 8. Where to start?

Of the three shot stats (for the games in which we have the data), Canada’s shots are the most consistent:

The standard deviation for Canada’s shots was 9.0% of the mean, the lowest of the three stats. We will therefore assume that in each of the final three games, Canada had an average number of shots for them: 37, since you cannot have partial shots in a game. It should be noted that the number of shots Canada had in a game bore no relationship on the number of goals they scored; the coefficient of correlation is 0.08, which indicates a random relationship between the two.

So now we have an estimate of Canada’s shots in each game. But what about the Soviets? Their shots varied much more than Canada’s did. To estimate the USSR’s shots, I’m going to use the average of two models, because of two very high correlations.

The coefficient of correlation between the Soviets’ shots and their goals is 0.82, which is very high, indicating a strong relationship. So I came up with a simple model to estimate shots based on the number of goals. For the first five games, the average error of this model is 23% of the mean, which is about as good as you’re going to get. The model is:

Shots = (Goals x 3) + 15

This model gives the Soviets 21 shots in Game 6, 24 in Game 7, and 30 in Game 8.

The coefficient of correlation between shot differential and goal differential for the first five games is 0.76, which also indicates a strong relationship. I came up with another simple model (with an average error of 34% of the mean, but whose total estimated shot differential for these games is off by only one) that uses goal differential to estimate shot differential, as follows:

Shot differential = (Goal differential x 2) +11

Which gives shot differentials of 13 for each of the last three games. And since we have estimated Canada to have 37 shots in each of these games, we estimate the USSR to have 24 shots in each of these games.

Taking the average estimates of the two models, we have the following shot estimates for the Soviets: 23 in game 6, 24 in Game 7 and 27 in Game Eight.

Taking it all together then, here is the estimated total performance for each of the three goaltenders who played in the Summit Series (with * marking estimated figures):

Vladislav Tretiak

Tony Esposito

Ken Dryden                          

So the fact that Ken Dryden gave up more goals than Tony Esposito in the same number of games may be partly due to his facing a few more shots. But all in all, Dryden did not play well, except probably in game 6. Note that I am not saying that Dryden was not a good goaltender. He just had a bad handful of games, and you cannot judge a player based on a handful of games. Recent research has determined that Dryden led the NHL in save percentage on at least one occasion. He also had the misfortune of playing game 1, when Team Canada was full of bravado, and simply assumed they would win all eight games. Taking out game 1, his estimated save percentage in .852, which is more respectable.