Hockey Outsider
Registered User
- Jan 16, 2005
- 9,438
- 15,595
This thread replaces an older post that I made on adjusted save percentage. (I stopped updating that one in 2009 - time flies). I won't repeat the methodology in detail, but essentially each goalie's save percentage is normalized to an environment where the league average is 90.5%. I also normalize the schedule to 82 games, and 30 shots per sixty minutes (otherwise certain seasons would count for more/less for reasons outside of a goalie's control). This is very close to the 2006-07 season.
Self-criticisms of this method:
Self-criticisms of this method:
- Save percentage is affected - perhaps significantly - by the quality of a goalie's team. Are they playing on a dynasty, protected by three Hall of Fame defenseman and a Selke trophy winner? Or are they playing on an expansion team, full of castaways? We now know that there's a significant difference in save percentage at even strength compared to while shorthanded - goalies on teams that are disciplined will have an advantage.
- Save percentage is, by definition, an average. Like points per game, it tends to decline over time. A goalie who retires at (or shortly after) his peak will have an advantage relative to others who play long, full careers. (This is why I also present a statistic for "best seven years", to try to remove this distortion).
- Save percentage doesn't take into account how important or "clutch" a save is. A stop in overtime and a stop during a 7-0 blowout count as the same.
- Save percentage doesn't take into account consistency. One goalie might stop 28 out of 30 shots every game, while another might allow four goals on half the nights, and get a shutout on the others. I suspect (but can't prove) that the first goalie would be more valuable over the long run, even if their save percentage is the same.
- Save percentage doesn't take into account goalie skills other than stopping the puck. For example, puckhandling is completely ignored. So are intangibles such as a team being willing to take more risks and (hopefully) generate more offense, since they have confidence in their goalie.
- As I'll discuss in more detail below, what's average changes significantly from year to year. For example, in 1959-60, 82% of the minutes played league-wide was from a Hall of Fame. In that year, multiple Hall of Famers were below average. Twenty years later (1979-80), only three Hall of Famers appeared in as little as 40 games. Due to expansion, the average quality of a goalie in 1979-80 was likely much lower than it was twenty years before - but both seasons are adjusted to the same scoring environment.
- I rarely add a disclaimer relating to one specific player, but in this case I think it's warranted. Terry Sawchuk's numbers are, frankly, a huge disappointment. But this excludes his five year prime, where he was the first- or second-team all-star five seasons in a row. Keep that in mind when reviewing his numbers.