A player performance management strategy that has come out of the NHL’s analytics movement is to avoid playing goalies two nights in a row whenever possible. The first popular article written on the topic was Eric Tulsky’s analysis that concluded that goalies should almost never play back-to-backs, with the difference between a tired and a rested goalie’s save percentage being 0.020. According to that study, a rested backup goalie was almost always a better option than a tired starting goalie. The significance of this result helped create a shift in goalie usage in the NHL, where teams are resting their goalies much more than in the past.
However, Tulsky’s analysis was limited to 2 seasons of data and was prone to sample bias. Similar studies where performed later, notably by Olivier Bouchard, and the results found a much smaller difference between tired and rested goalie. Bouchard found that the difference in save percentage over 7 seasons was 0.002, which has much different implications for player performance management strategy.
Analysis of QMJHL goalies
I decided to perform a similar type of analysis on QMJHL goalie data, although with a more vigorous process to avoid any biases or misleading results. To ensure that I had a large enough sample size, I captured goalie data from every regular season game from the 2010-2011 season to the 2016-2017 season. To avoid any effects caused by team fatigue, I filtered down my data to only include the 2nd games of back-to-backs. Finally, I only used data from goalies that had played at least 4 back-to-back games tired and 4 back-to-back games rested. At the end, I had 84 “goalie seasons” to compare the impact of playing tired versus rested.
The result: an average difference in save percentage of 0.002
This matches what Oliver Bouchard’s study of NHL goalies had found. It is interesting that despite the differences in age, skill level of shooters, and season length, the impact is essentially the same. Going back to the initial question, should goalies play back-to-back games in the Q? Based on the numbers, there isn’t much of an effect. The only real situation where this finding could be useful would be if a team has two very evenly talented goalies. For example, Charlottetown has had essentially two starting goalies this year:
- Mark Grametbauer: 0.897 SV% in 40 games
- Matthew Welsh: 0.899 SV% in 38 games
Given how closely those 2 goalies have performed this year, the numbers would support the goalies splitting back-to-backs. However, the fatigue effect is minimal so other considerations like recent form, injury proneness, and big game experience could be more important. Overall, teams should be expecting a slight decline in a goalie’s performance in the 2nd game of a back-to-back.