Predicting Vezina: The Real Rules for Trophy Voting
Ottawa Senators goalie Andrew Hammond (aka the “Hamburglar”) has won 90 percent of the games he’s started this season. He stops 95.6 percent of the shots he faces, and has a 1.35 goals-against average.
These numbers make even Carey Price’s stats look sad and weak. But even if Hammond keeps up this incredible pace for the rest of the season and launches Ottawa into the playoffs one thing is certain: he will not win the Vezina Trophy.
Nothing in the NHL’s description of the award would prevent him, of course. Since the 1981-82 season, the Vezina has gone to “the goalkeeper adjudged to be the best at this position as voted by the general managers of all NHL clubs.” No further guidelines are provided for the award, meaning GMs are free to vote according to whatever criteria they choose. Wins? Sure. Goals-against average? No problem. Shutouts? That’s fine. Save percentage? Seems reasonable. Reputation? Yeah, he’s earned it. Dice roll? Nothing against it.
Despite the surprisingly high number of NHL GMs who seem to make decisions based on whim or astrology, over the last 30 years, stable voting criteria have emerged. Some of these are unofficial rules that have never or seldom been broken. Others are strong guidelines that prioritize particular statistics.
Rules of Exclusion
- You must play in at least half your team’s games.
Even though any goalie who plays even a single game is eligible to win, the GMs have set the bar much higher. No goaltender playing fewer than 41 games, or 2400 minutes, has ever won the silverware. This number represents half a team’s games, and ensures that keepers are being judged on a significant, sustainable body of work.
The spirit of this rule is so strong that even when a goalie does hit the minimum, he’s not guaranteed consideration. In the shortened 2012-2013 season, Ottawa Senators’ starter Craig Anderson led the league in both save percentage and goals-against average. He played 25 games, more than half the 48-game total for that season, but it’s clear the GMs needed to see him sustain his play over a more significant stretch of games. He wasn’t even nominated.
If Anderson wasn’t recognized because he played only 25 of 48 games, there is no chance the Hamburglar will do better playing 25 of 82. Sorry, Ottawa fans.
- You must have a winning record.
Rule one showcased the reason and knowledge that can emerge when experienced people come together to make a decision. Rule two, however, represents a baffling bias so deeply rooted that we may never see it eradicated. It’s understandable that general managers would place disproportionately high value on victories (their jobs depend on them), but does anyone else actually believe that a losing season couldn’t be had by the best goaltender? Roberto Luongo never won a Vezina in his prime because his incredible play always came with a terrible record.
Dominic Hasek won six Vezinas in eight years. The two he didn’t win in this stretch support the consistency of both rules: in the first (1995-1996), he had a losing record. In the other (1999-2000), he played too few games.
- Your team should make the playoffs.
This rule, closely related to number two above, was unbroken until Sergei Bobrovsky won in the shortened 2012-2013 season. His Columbus Blue Jackets just missed the post season despite his impressive 21-11-6 record, but he was dominant enough behind a poor team to break 30 years of voting tradition and win. Had Bobrovsky played at the same level, with the same overall numbers, but lost more games than he had won, there’s no doubt that the Vezina would have remained out of reach.
Rules of Inclusion
- Three stats matter most: save percentage, goals against average, and wins
Since 1983-84, when save percentage was first tracked, it’s been the biggest predictor of Vezina success: 60% of winners lead the league (among goalies who played over 2400 minutes). A third of the winners led the league in goals against average, but in all but one case, that goalie also led the league in save percentage. Wins predict about 27 percent of Vezina winners, and there is very little overlap between this and the other two measures.
This data tells us that goals-against average is an essentially useless evaluation tool because of its overlap with save percentage. Wins tells us something interestingly different; I’d argue that Vezinas awarded based primarily on reputation use wins as the key support.
- Goals Saved Above Average is actually all you need to know.
GMs may not realize this, but all their hard work evaluating goaltenders’ statistics is no longer necessary. A single number, Goals Saved Above Average (GSAA), is the most predictive measure of Vezina victory.
A goaltender’s GSAA shows how many additional goals a league-average keeper would have allowed in his place. Historically, it does a slightly better job than save percentage alone (63.3 percent vs 60 percent), but since the 2004-2005 lockout, it’s surprisingly more accurate. A hefty 78 percent of winners have finished first in GSAA since then, with 100 percent finishing either first or second. Only 56 percent of the winners have finished first in save percentage, with 78 percent finishing first or second.
The predictive power of GSAA comes from its combination of the rules listed above. A goalie’s workload is a key component of GSAA: keepers who play more have the potential to record a higher score. Save percentage comprises the other half of the GSAA measure, and as we have seen, it’s the traditional statistic GMs take most seriously.
Don’t believe me that such a simple measure could capture all the complex evaluation that goes into picking the league’s best goalie? Try this magic trick. Using all the criteria I’ve outlined above, choose three Vezina nominees, and pick the winner. Then, go to Hockey-Reference and sort by GSAA.
I’m going to guarantee your choices are the first three or four names listed, likely in the same order. Let me know how this worked for you in the comments below.
What is, versus what’s right
Everything I’ve outlined here involves who has been, and who will be, chosen for goaltending’s highest honour. However, determining what is, and why it is, is very different from considering how things should be. Is workload with raw save percentage the best measure of a goaltending season?
The answer is no; explaining why this is the case will have to wait until next time.
~ Paul Campbell teaches in the humanities program at Wilfred Laurier University, and also writes for thehockeychat.com. He’s a former CIS goaltender and women’s team goaltending coach at Mount Allison University. Originally from Cape Breton, Paul now lives in Guelph where he’s conducting experiments to determine which of his sons should don the pads (once they learn to stand on skates). Respectful feedback and spirited discussion are always welcome in the comments below.