I have this theory about hockey that I’ve been advocating on Twitter and in personal conversations, after doing a lot of disorganized private research. But I’ve never put it down in print before. The theory goes like this: goalies don’t improve. Actually, I prefer to capitalize it, in order to impart the awful character of divine law. Goalies Don’t Improve.
It is a little more complicated than that, of course, but it is a merit in a controversial hypothesis to have a simple, shout-able basic form. What I’m really saying is that that NHL goaltenders approach their peak in quality, relative to the league, earlier than almost anybody thinks. And that in the overwhelming majority of cases, waiting around for a young goaltender who has established mere adequateness to advance and become an all-star contender is probably a waste of time.
Goalies who will ultimately be very good almost always prove it pretty early in their careers. But the converse is obviously not true: many goalies who make an impact at a young age are out of the NHL, or headed that way, by the time they are 30. (Anybody remember Jim Carey, Net Detective? He’s two years younger than Martin Brodeur! Still!) Analyzing goaltenders is difficult for subtle statistical reasons, but it is important to think clearly about the subject. With the Sochi 2014 Winter Olympics approaching, Canadian fans have been alerted to our country’s relative weakness at the position—partly because some young prospects Didn’t Improve.
There are a few threads of evidence pointing towards the GDI hypothesis. But even rough formal study supports it. If you take the last 30 years or so of NHL save-percentage data, and just figure out everybody’s save percentage by age, you’ll be surprised by the early peak. Collectively, goalies seem to max out between the ages of 21 and 23 and then drop off more or less monotonically. (A chart compiled by stats whiz Gabe Desjardins got me thinking about this.)
This is true even though overall league save percentages have been climbing by close to a point a year for a while now. Goalies lose ground even faster if you adjust for that, pulling the performance-versus-age curve down on the right; and the GDI effect is still present when you compare goalies to their personal peak or their career-mean performance.
If you’re just a observer of the game who doesn’t dig math, I think that years of experience watching the NHL teaches the same thing. Many of the great goalies of the last three decades are guys who devastated the league as young men, like Martin Brodeur or Patrick Roy or Roberto Luongo. Some are players who were awesome somewhere else but got a late start in the NHL, like Dominik Hasek or Tim Thomas. (Almost as soon as some pro club gave Thomas a starting job, he was overall MVP of the top Finnish league—at a time when his Finn rivals were collectively preparing a mass invasion of the NHL.)
But which goalies fit the description “Started out meh, eventually became elite”? Someone on Twitter recently defied the GDI effect successfully, giving me an example I couldn’t explain away: Tomáš Vokoun, who was sincerely average through 2,500+ NHL shots and then levelled up like a video-game character. So it can happen.
But it’s rare. If there is a second example, it’s Dwayne Roloson. We know exactly how Roli did it: by radically reinventing his style, breaking down his whole method of playing goal and recreating it, in Minnesota. Roloson is an unusual, extremely intelligent man. His case may testify to the difficulty—as much as it does to the possibility—of such a thing.
If you pay attention, you will often hear sportswriters and hockey broadcasters assure you that goalies peak late and need time to “learn” the NHL. Every day they name some struggling “young” 25-year-old who might still leap to stardom. This sort of talk, I now think, is mostly nonsense. The position of hockey goalie really is one-dimensional compared to the others, and a 25-year-old should be conceived of as already facing age pressure. Reaction times, vision and flexibility are particularly important in goaltending, and they degrade early. The now-orthodox butterfly style destroys hips and knees, and it is not uncommon for junior goalies to have already logged enough hours to require surgery.
Experts have a powerful incentive to insist that being a goalie involves a panoply of learned arcana only available to professionals. But general managers (and national-team selectors) might want to be wary of goalies who have reached unrestricted-free-agent age (typically 27). Indeed, some statheads have already reached this conclusion on the ugly evidence of the contracts themselves, without studying the aging curve as such. Goalies Don’t Improve. Tell your friends.