David Hutchison | Jan 22, 2019 | 0
Vezina Race: What Do The Numbers Say?
Now that the season is over, it’s time to hand out the awards. This year’s Vezina race is a close one, with a number of players working their way into the conversation. There will be a lot of statistics tossed around this year, sometimes haphazardly. This is a good time to take a closer look at them. Overall, the numbers indicate that the most well-rounded season isn’t that of popular Vezina frontrunner Braden Holtby or Rangers hero Henrik Lundqvist at all.
Statistics are, of course, only part of deciding who deserves to be “adjudged the best at his position,” in the words of the award criteria. The General Managers, when voting for the award, tend to use criteria such as overall save percentage and GAA, but these aren’t really the most informative statistics we have available to us. In debating who should get the award (as opposed to predicitng who will get it), I’d argue for using a different set of measures.
So what statistics are the best to look at?
1. 5v5 save percentage – This covers the majority of the time a goaltender sees in a season and gives a broad starting point for comparing seasons to each other. It is preferable to overall save percentage because it compares apples to apples in terms of manpower.
2. PK save percentage – Still, penalty killing is a critical skill for a goalie. Some players who do well at 5v5 don’t do well on the penalty kill, where luck may have a greater impact or where their weaknesses may be more exposed. However, since every goaltender sees a different amount of time shorthanded, it’s best to separate it out for evaluation, rather than combining it with other situations. Again, apples to apples as much as possible.
3. High Danger save percentage – “High Danger” shots are shots from the slot plus some rebound and rush shots not taken from the slot. These are known to have a lower save rate league-wide and HD save percentage makes up most of the difference between goalies in the NHL. Used in conjunction with 5v5 save percentage, HD save percentage helps to differentiate between an elite season and a good one.
4. Adjusted GSAA/60 (also called Mercad/60) – GSAA is Goals Saved Above Average. It compares the goals a goalie saved to the number of goals that would have been saved at league average save percentage on the same volume of shots. Hockey-reference.com collects this stat but doesn’t separate out either 5v5 time or take into account the amount of time a goalie has played.
Adjusted GSAA uses location data in the NHL play by play files to determine how many or fewer goals a goalie has saved than a league average save percentage on the same volume of shots in each of three danger zones–high, medium, and low. In other words, adjusted GSAA applies save percentage in each zone to shot volume in each zone to get a “better/worse than average” number.
5. Adjusted Save Percentage – Found at war-on-ice.com, Adjusted Save Percentage applies this location adjustment to a goalie’s 5v5 save percentage to get a sort of normalized save percentage. While it is generally considered an improvement over 5v5 save percentage, my concerns about it have been growing recently. I don’t think it tells us what we have been assuming it does and I will thus weight it much less heavily than others will. But for the time being, it is part of the equation and so I present the data here.
After pulling the data from war-on-ice.com (or via Nick Mercadante on twitter in the case of AdjGSAA/60), I made a cutoff of 50 games played, which is a minimum for being considered for a season-long goaltending award. That gave me 20 goalies to consider. I ranked them by each of the above statistics. Then I simply weighted each player’s rank by the following formula. Lower scores are better.
(5v5 sv% + HD sv% + AdjGSAA/60) + (2 x PK sv%) + (3 x Adj sv%)
Thus goalies who ranked better in the most important measures were rewarded more than those who ranked better in the less important measures.
This table shows the ranking in each stat for every goalie who made the top ten in any of the measures. [For AdjGSAA/60, ranking is among those goalies who faced the most shots, above around 1000 shots at 5v5, or those who show up in blue in the graph found here.]
|5v5 Sv%||PK Sv%||HD Sv%||AdjGSAA/60||Adj Sv%||Rating|
Notably, although Corey Crawford’s name has floated around edges of the Vezina conversation for a little while, this method ranks him and Marc-Andre Fleury higher than most other analysts do. Ben Bishop is ranked third mostly on the strength of his penalty killing results, while Henrik Lundqvist and Steven Mason are both dropped from the top ranks because of their HD and PK results. A stronger weighting for Adjusted Save Percentage, making it equal to the top three measures, would not change the top four goalies, but Schneider would jump to fifth and Holtby to sixth.
Of course, none of this predicts who actually will win. It just shows that when looked at using the set of measures that best capture all aspects of a goaltender’s season, Crawford and Fleury have the strongest statistical cases for the award.