The Shutout Paradox
Consistency isn't always implied by excellence.
Thought experiment:
Say you have two goalies. Going into a season you know the following:
goalie X will have started (and completed) 75% of all games (without ever being played in relief) and finish with a .900 sv%, a 2.50 GAA, and 5 shutouts
goalie Y will have played the same number of games, starts, minutes, etc. as goalie X and finish with a .900 sv%, a 2.50 GAA, and 0 shutouts
teams X and Y (who goalies X and Y play for) are roughly equal in effectively every area of play and finish with an identical record.
Furthermore, let’s say that the play of the team X is not a factor in Goalie X’s shutouts, nor is team Y the reason for Goalie Y’s lack thereof.
In fact, team X wins all five of goalie X’s shutouts by a score of at least 2-0, and Team Y does not lose any games 1-0 all year.
All this to say, would you rather have Goalie X’s 5 shutouts, knowing the fact that since Goalie Y reached the same GAA without posting 5 shutouts — and since all other factors are equal— thus played more consistently over the other games?
‘Who cares?’
Well, look at it this way. Let’s say teams X and Y play in a league with an 80-game schedule: that means that goalies X and Y over the whole season
each played 60 games
allowed 150 goals
faced 1500 shots
and faced 25 shots/game on average.
Goalie X had a 2.50 GAA, .900 sv% over 60 starts, yes, but in 55 non-shutout games, Goalie X had a 2.72 GAA and a .890 sv%.
‘Are you saying that shutouts are bad?’
No! What?! Who would even say that?1
Shutouts aren’t bad. They’re extremely good outcomes. Shutouts are guaranteed non-loses at all levels of hockey. The shutout paradox only applies in cases where all other stats are equal. Even still, there comes a certain threshold where consistency is more valuable — and a certain threshold where it’s not. The longer the time-frame, the more valuable consistency is. The astute among you might note that the difference between a .900 sv% and a .890 sv% is approximately the difference between stopping 9/10 shots and 8/9 shots. Over the 55 non-shutout games, this would result in goalie X allowing about 15 more goals than goalie Y. But does it matter? If those 15 goals are diffused are scored in dead time, it won’t. But if they’re in a close game they do. But remember team X and team Y finish with identical records, so for our purposes it’s inconsequential.
It’s important to always consider consistency Quality starts and Really Bad Starts are good ways of doing this. And while it is most often the case that shutouts are indicative of outstanding performance, it’s important as well to note how often a goalie gives their team a chance to win — not just guarantee them.
Author’s note: It’s me. The answer is me. I’m the guy who wrote that.