Quantifying Carey Price: Has He Earned the Hart?
2015 will be known as the year of the goalie.
Three of the year’s biggest individual stories are Carey Price’s dominant season for Montreal, Devan Dubnyk’s unlikely comeback in Minnesota, and Andrew Hammond’s impossible run to push Ottawa into the playoffs. More generally, the league average save percentage of 0.915 is the highest in history; a more even distribution of ice time has led to fewer skaters reaching historically noteworthy scoring benchmarks; and finally, never has so much attention been paid to goaltender statistics and evaluation.
If there were ever a time for a goaltender to win the Hart trophy, this is it. Despite the unprecedentedly goalie-friendly environment, however, only one has a chance at winning, and that has to be Carey Price.
The Three-part Argument
Last time, I proposed a three-pronged method for determining whether a goaltender should win the Hart trophy. It’s worth reviewing because it lays the theoretical groundwork for the case I’m making here.
The central point is that a goaltender should only be chosen for the award if he shines through three kinds of comparison: 1) A within-position comparison; 2) A comparison to historical precedents; 3) An inter-position comparison.
The first is essentially the argument that the goaltender should win the Vezina. The second aims to show the goaltender’s season has been historically significant, and comparable to other goalie Hart winners (or near winners). The third (and trickiest) shows his superiority relative to skaters in his own season.
- First Period: The Vezina
Although I’m not alone in believing that Price is the most technically proficient NHL goaltender, observation shouldn’t do more than direct closer attention to a given candidate – we all have our favorites, and for reasons that have little to do with objective performance measures.
Looking at the Big Four traditional goalie stats, Price leads in three: His .933 save percentage, 1.96 goals-against average, and 44 wins all top the league, while his nine shutouts are good enough for second place. This will be enough to win him the Vezina on its own, but because all of these statistics are essentially save percentage with various team effects added, we have to go deeper.
The best predictor of Vezina success, Goals Saved Above Average, is a runaway win for Price: the gap between his 36.7 additional goals saved and second-place Dubnyk’s 23.7 is the same as the separation between Dubnyk and twelfth place.
Looking beyond raw save percentage and workload, we get an even more detailed picture of Price’s quality.
Price’s even-strength save percentage (less susceptible to the influence of team systems) ranks second behind Philadelphia’s quietly effective Steve Mason. Price leads the league in adjusted save percentage, which takes into account the danger of a given shot’s location. His high danger save percentage, the component of adjusted save percentage found to best differentiate elite from average goaltending, sees him in fifth place, just behind New Jersey’s Corey Schneider and ahead of Dubnyk.
(All rankings only include goalies playing more than 2,400 minutes)
In short, no goaltender rates nearly so highly on every measure as Price. Dubnyk, Schneider, and Rinne are all playing in Price’s approximate league, but while Price’s numbers are not surpassingly dominant in most categories, they form a far brighter constellation than anyone else’s.
- Second Period: Historical Standing
The historical significance of Price’s raw save percentage has been lauded; his .9334 gives him the sixth highest total of all time, which seems to make his historical case convincing. It puts him above Jose Theodore’s Hart-winning season, and above one of Dominic Hasek’s. However, this comparison’s simplicity renders it frankly dishonest.
This season’s record league-average save percentage actually makes Price’s achievement less impressive.
To adjust for era, Sportsnet’s Stephen Burtch determined the shot-weighted league-average save percentage for all goaltenders with more than 20 games played for every season since 1982. The relative magnitude of a goaltender’s performance above his season’s average is a far better measure of an outstanding season than raw save percentage. Hasek’s 1994-95 season tops the list at an incredible 2.56 standard deviations above the mean (z-score).
Theodore’s 2001-2002 Hart season places second at 2.55. Hasek’s Hart-winning 1997-98 and 1998-99 seasons come in fourth (2.51) and fifth (2.47).
By season’s end, Price’s z-score of 1.66 placed him well outside the top 20 seasons of the last three decades. To give this number a bit more context, a score over two indicates a highly significant separation from the average. Andrew Hammond’s score of 2.33, were his season long enough to qualify, would place him, impressively, in the top eight. Price’s relatively low value indicates a good season, but nothing statistically, historically beyond the probable.
One could argue that Price’s leading three of the Big Four goaltending categories is a rare accomplishment (it is), as is setting a new franchise record for wins (it is, too). Unfortunately, these feats are so highly team dependent that they serve as poor markers of historical significance for an individual goaltending season.
Price is the best in the game, but his relative performance doesn’t approach the historical greatness of previous goalie Hart winners. In any other year, Price’s season would not have merited the Hart.
- Third Period: Goalies vs Skaters
The most difficult comparison in hockey is between goalies and anyone else. Because there is essentially no overlap between the skills needed to be a prolific scorer and an impenetrable goaltender, we have to consider their contribution to either side of the most basic common metrics: goals and standings points.
War-on-ice.com features a new Goals Above Replacement measure that estimates the goal contribution a given player has made above replacement level. Burtch uses this measure (which Carey Price lead) to make his own Hart case, but my initial worry is that goaltenders may be likely to finish first most years. The presence of Price and Schneider in the top three suggests that, while the assigned value may accurately represent a player’s contribution, an adjustment would have to be performed to compare the relative value of the most elite skaters and goalies. Because the application is still being developed, performing such an adjustment is not yet a straightforward matter.
While we await the continued development of Goals (and Wins) above replacement, we can look to hockey-reference.com’s Point Shares statistic to make our impossible comparisons more manageable. By estimating a player’s overall impact on his team’s standings points, Points Shares seems tailor made to determine Hart worthiness; what matters more to your team’s success, after all, than standings points?
One major impediment to the straightforward use of Points Shares for Hart evaluation is the high value it attributes to goaltending. Again, this may be an accurate reflection of the position’s high importance, but it means that a goaltender would win almost every year. In 2015 for example, goalies would finish first through fifth, with the top skater tying for fifth place. Taking all goaltenders into account, their average Points Share is 4.79, while the average of all skaters is 2.56.
Part of the problem with this comparison is the sheer number of skaters who lace up in a given season, compared to the number of goalies: 92 goalies saw action in 2014-15, compared to 882 skaters.
Most of the goaltenders appeared in over 20 games, while many of the skaters played just a few total minutes (regardless of the number of games in which they appeared). Such a mass of non-contributors decreases the average significantly, and subsequently, makes any legitimate contributor look far superior.
To focus on the skaters who had the potential to make meaningful contributions, I narrowed my consideration to those playing more than 1230 minutes, or 15 minutes per game over 82 games. This gave a much more commensurate Points Share average (5.77) compared to all goaltenders.
Then, for goalies and skaters separately, I determined their relative distance from the mean using z-scores, as above. The results, when combined, are interesting:
|1. Alex Ovechkin||2.79|
|2. PK Subban||2.71|
|3. Carey Price||2.53|
|4. Erik Karlsson||2.51|
|5. Rick Nash||2.30|
A single goaltender appears in the top 5, which is more in line with what we might want or expect from a metric used to choose Hart candidates. Another benefit is that top defensemen (historically even more invisible to voters than goalies) are highly rated. This makes even more sense when we consider the relatively low individual point totals among elite scorers this year.
Despite, or perhaps because of the remarkable success of this season’s goaltenders, Ovechkin’s 53 goals sets him much further above his peers than Price. Adjusting for era, Ovenckin’s total is the fifth best since 2000 (he also holds first and fourth place on that list), and the sixth best since 1995.
In the end, while it’s clear that Price should win the Vezina, his season was neither sufficiently historically significant nor far enough above his skating peers to warrant the Hart. I can’t help hoping he wins, and I think he will, but the bar for a goaltender to be the league’s MVP has to be set higher at a time of increasing save percentages and decreasing 40 and 50 goals seasons.
I might lose my goalie union card for saying this, but I’m punching my (imaginary) ticket for Ovechkin.