Red Wings 2015-16 Season Preview: The Goalies, Part II of II
Detroit's goaltending last season, on the whole, wasn't as bad as it sometimes seemed to be, especially late in the year. It just wasn't any better than what you could get from a perfectly league average goalie on most nights.
5v5 adjGSAA/60 is a stat that shows how much better or worse the goalie in question (Howard and Mrazek) performed with their shot loads from low, medium and high danger zones, than had a goalie with perfectly league average sv% received the same shots. The end result is a 60 minute rate statistic that shows how many more or fewer goals a goalie saves per game than if a league average goalie were in his place. To learn more about this stat and its advantages for comparing goalies, read this piece from Blueshirt Banter
As you can see from the chart, both goaltenders underwhelmed on the season. Mrazek basically performed at league average, and Howard was below. Considering that the two encompassed nearly every minute of goaltending for Detroit last season, this means that Detroit got pretty much league average goaltending game in and game out.
Mrazek and Howard did run cold at about the same point towards the end of the season, but both were relatively close to league average for the season.
Howard's career peaked between ages 27-29. As he has played 338 NHL games to this point, and turned 31 this year, it should be safe to say that the likelihood that he will reach the heady days of the lockout 2012-13 season, where he was saving a whopping half goal more than league average per game, are a few years behind him now.
Howard's performance going forward is likely somewhere in-between, as indicated by his rolling average. Performance slightly above league average would put him squarely in the top 20 or so goaltenders amongst those with 1200+ minutes, and that is pretty reliable starting goaltending for a team that can defend adequately and who have a top 6 forward group that can fill the stat sheet. If I weren't spoiled by The King, I would take that on my team, though the price paid in terms of AAV is exorbitant.
Mrazek is a bit of a question mark. He did nothing to distinguish himself last season, and his performance gradually declined. Many fans may feel that he came out hot and then fell off a cliff:
While, his 10 game rolling 5v5 adj sv% above may tell support that opinion, his game by game adjGSAA/60 and the linear regression overlaying it shows that it probably was a bit more of a gradual decline to what may end up being his expected performance:
The fact that adjGSAA/60 doesn't show the same kind of cliff indicates that his defense may have been letting him down more towards the end of the season. But he did slide down a gradual hill in terms of performance. So he isn't immune to questioning. But ultimately, his is a very small sample size. So it is hard to say at this point with any real degree of certainty what to expect going forward.
Goaltenders generally need seasoning and I tend to think that the truth about performance that starts off well-above or well-below league average is probably somewhere in between (mind blowing, I know). Mrazek has performed admirably enough. But I would not consider what he has done thus far in his short career to be enough to warrant regular starting consideration over a guy who only 3 years ago led the NHL in adjGSAA/60, regardless of recent struggles.
Does this clear anything up for you Detroit fans? Not really. Goalies...are...voodoo.
That's a tough act to follow, so I'm just going to add a few things some readers and commenters asked about. I'll also be doing fairly periodic goaltending analytics updates throughout the season, but today's exercise will consist of a very short, simple, and hypothetical FAQ.
What is adjusted save percentage, and which goalie's was better?
Adjusted save percentage is a statistics that attempts to take shot quality into account by separating shots into "danger zones." Adjusted save percentage is rapidly approaching mainstream status, and you can read all about how it works in this blog post at War-On-Ice.
Regarding Red Wings goaltending performance from last sesason, Petr Mrazek exceeded Jimmy Howard in both 5v5 adjusted and traditional 5v5 save percentages. While the smaller gap between the two players' performances in adjusted save percentage suggest Howard was on the receiving end of some tougher luck than Mrazek, the rookie goaltender's performance was still superior in all three danger zones than that of the veteran's. It is worth noting that Howard faced almost a full shot more per 60 minutes than Mrazek.
How did the goalies' performances compare prior to Howard's injury?
This was an answer I found incredibily interesting, albeit with an even more pronounced sample size caveat than the rest of this exercise filled with them. While shots against per 60 and traditional 5v5 save percentage were both quite similar, Mrazek bested Howard by over a full percentage point in 5v5 adjusted save percentage. The biggest difference between the two was Mrazek's 2.37% advantage in the medium danger zone.
What does it all mean?
There is still much work to be done in regards to the danger zones in adjusted save percentage. How much of it is definitive? Which zones contain more noise in their results than others? Some of the smartest minds in hockey analytics still disagree on those answers. When I reached out to Nick for his assistance on this project, I gave him my angle on the entire situation: we know what we have in Jimmy; Mrazek is still a mystery, but one who is probably more valuable going forward. As Nick noted, Howard's past performance warrants that he receive playing time. But he also used the word "exorbitant" in describing the salary Howard is being paid to produce around a league-average level. In the end, Nick agreed with my assessment. And I agree with his: goalies are indeed voodoo.