On the surface, the Devan Dubnyk narrative is simple.
From 2014-19, Dubnyk was a top goaltender — second only to Braden Holtby in Games Started, fifth with a .920 Save % (200-plus games played), a positive Goals Saved Above Average (GSAA) in every season, third place in 2015 Vezina Trophy voting and fifth in 2017.
That would make Dubnyk’s 2019-20 — .890 Save %, -16.23 GSAA, lost the number-one job to Alex Stalock — an obvious aberration. Dubnyk’s wife also underwent serious health issues during the year.
With that in mind, who wouldn’t want to trade for Dubnyk, especially at just one year left and a $4.33 million cap hit? Naturally, the San Jose Sharks, looking to supplement Martin Jones, have been linked to Dubnyk for the better part of a week — and it looks like a deal is getting close.
Digging deeper, however, advanced stats suggest that Dubnyk has had more than one bad year.
That .920 Save % from 2014-19? That positive GSAA from each season? Dubnyk was just doing what was expected.
GSAA is roughly a goaltender’s Save % versus league-average Save % — applied to the number of shots that the goaltender has faced. However, it doesn’t account for shot quality or the team in front of the goalie. So a netminder behind a strong defensive squad is less likely to face high-danger scoring chances — therefore, his Save % and GSAA will likely be better than the average.
But there’s another stat that suggests Dubnyk’s Save % and GSAA should’ve been higher — not just this year, but for the last five seasons.
Per Evolving-Hockey’s Goals Saved Above Expected (GSAx), Dubnyk has allowed more goals, in all situations, than he should have for the last five years.
To be exact, per this model, Dubnyk surrendered 27.49 goals more than an average goaltender would’ve this year. By this measure, he was the worst keeper in the NHL this year, worst than two-win Jimmy Howard.
GSAx measures how many goals that a netminder has surrendered above or below expected, based on the shot quality faced. Public measures of shot quality are mostly based on shot location.
In translation, according to GSAx, Dubnyk has lived off a staunch Wild defense — and he hasn’t outperformed it since 2014-15.
So should the San Jose Sharks be concerned about Dubnyk’s GSAx? Is he really the right guy for a not-as-tight Sharks defense?
Goals Saved Above Expected
First, let’s get this out of the way. I’m pretty sure that the San Jose Sharks are aware of Dubnyk’s less-than-stellar numbers in this regard over the last five seasons. This (fantastic) modeling from Evolving-Hockey is based on publicly-available data and costs (well worth it) $5 a month.
Independent of each other, Silverman and Anderson agreed on one thing that might have affected Dubnyk’s GSAx.
“I’m convinced they overused him for a while there,” Silverman wrote, “and it tanked his numbers for stretches.”
Anderson offered: “It doesn’t account for how much the goalie is used.”
As noted, Dubnyk was second in the league in Games Started from 2014-19.
Let’s also talk about how GSAx measures shot quality. GSAx pulls from the league’s data. Therefore, GSAx’s account of scoring chances is mainly shot location. The closer the shot, the more dangerous it is, the farther the shot, the less dangerous it is. Nothing else: Pre-shot movement (i.e. passing), odd-man rushes, shot velocity, and traffic are among the keys to goal scoring that are not accounted for by GSAx.
Anderson countered: “GSAx is based on the output of models based on league’s data — the league provides raw data, but not the modelled or ‘insightful’ pieces. It is mainly shot location and type but also has the game state like power play, shooter, and we can also infer some time-based variables, like if the shot is a rebound, or if the shot comes seconds after an event in the neutral zone, likely off the rush.
“But what you listed is all true and the biggest missing pieces, although those aren’t as important as people might think. For example, a team will often pass to get good shot locations, or elect to use a 2-on-1 to get a shot tight to the net, so shot location is a good proxy for those things in this model.”
All that said, Anderson added of the stat: “It isn’t by itself incredibly repeatable — which is good news for Dubnyk — and the public version relies on modeling on NHL data, which is somewhat incomplete and has some issues.”
He suggested of the league’s data: “It’s possible Minnesota specifically might have recorded shots further from the net which would make Dubnyk’s job look easier than it was in reality in half his games.”
Now all this doesn’t account for Dubnyk’s lost 2019-20 campaign — and once again, the 34-year-old netminder had plenty of reasons to not have focus — but they might prop up the more marginal results from 2015-19.
What Might San Jose Sharks See in Dubnyk?
Silverman pointed out that Adam Francilia, a Sharks goaltending consultant, has worked with Dubnyk for years. Nabokov authorized Francilia to work with Aaron Dell this year.
As Pierre LeBrun noted, San Jose wants to make sure that Dubnyk is comfortable with moving to West Coast, considering his wife’s health problems — and giving Dubnyk the chance to work more with Francilia could certainly be a lure. You want guys who want to play for you.
Doug Wilson also spelled out what he was looking for in a new goaltender on Friday: “Ideally — and it depends on what the cost is, the acquisition cost — is getting a guy who’s been a number-one, that’s a veteran, who wants to come in and compete for a spot.”
This sounds like Dubnyk — and not like Aaron Dell, for example.
So Dubnyk is a good fit for the Sharks goaltending depth chart, which at the moment, reads Martin Jones, then…21-year-old prospect Alexei Melnichuk?
“I’m not yet convinced that Melnichuk is ready for the big leagues and I don’t think anyone is,” Silverman said, “but in theory, they’re hoping that he’ll be ready sooner rather than later and would want a stopgap who has worked with Francilia before.”
But why Dubnyk in particular? The jury is still out on that and it will be fascinating to pick Wilson and Nabokov’s brains on the subject when the time comes.
“[Dubnyk] obviously has size and can manage the game a little deeper in the crease. Goaltender results can be heavily influenced by their environment — like any position, really — so if Nabokov or the rest of the staff like the skill-set, I am certainly open-minded,” Anderson conjectured. “Jones and Dell specifically were more aggressive than average, which may not have been a good fit for San Jose.”
Sheng’s Travel Fund
Help fund Sheng's travel! Every dollar goes to the cost of getting to and from Sharks road games.