I was inspired to write this article by a post over at Behind the Net where they discussed goalie even strength save percentage by age. In that article they came away with the conclusion:
Analyzing the data every way I can think up, there is no evidence whatsoever of any relationship between goalie age and even strength save percentage.
But, as I pointed out in the comments there are two serious flaws with their analysis. The first is that they do a linear regression analysis which is probably not the best tool to use as one can reasonably assume that a goalies skills do not progress linearly as he ages. One would expect a goalie would improve early in his career, plateau for a few years, and then regress in the latter years of his career. A hypothetical example might see a goalie have the following save percentages for his 10 year career: .900, .905, .910, .915, .920, .920, .915, .910, .905, .900. Fitting that data to a linear equation would net a slope of zero indicating no relationship between age and save percentage. Clearly from the data though there is probably a connection between age and save percentage.
The second problem I pointed out in the comments was in the second part of the analysis where they just looked at goalies with 3 years or more of data. The problem with this is that one can probably expect the difference in a goalies skill level at age 26 and at age 28 to be minimal, if anything at all. Skill level variation by age is likely to take a lot more than 3 years to identify. The reality is if a goalie only played in the NHL from age 26 to 28 that was probably the peak of his career and he simply wasn’t good enough to make the NHL to have a save percentage to use before and after that point, but since only NHL save percentage was used we can’t know how the goalies skill level might have improved before age 26 or tailed off after age 28.
To fix these issues one really should only look at goalies where you have a lot of data over many years and not look at it using linear regression which isn’t going to work. So what I have done is identify all of the goalies that have played in the NHL for at least 14 seasons and whose careers continued at least into the 2000-2001 season. The list of goalies that meet these qualifications are: Curtis Joseph, Tom Barasso, Sean Burke, Ed Belfour, Dominik Hasek, Olaf Kolzig, Jocelyn Thibault, Chris Osgood, Martin Brodeur, Patrick Roy, John Vanbiesbrouk, Mike Vernon, Ron Tugnutt, Jeff Hackett, Craig Billington, Glen Healey, Kirk McLean, Mike Richter and Chris Tererri. That list includes quite a variety of goaltenders including hall of famers and career backups and everything in between. I then took averaged their save percentages by age and came up with the following chart.
From that chart there appears to be a very strong relationship between age and save percentage. Before age 22 and after age 40 there are only a small number of data points so the chart goes a little wonky at the edges but from age 22 to age 30 there is a fairly steady improvement in a goalies save percentage. Between the ages of 30 and 34 a goalies save percentage plateaus before starting a steady decline until age 40. This is very much as one would expect and it only takes looking at the data in the proper way to confirm our expectations.