The Good, The Bad, and The Ugly Regression

As a wise woman once told me, it’s inevitable and “regression eventually comes for us all.” This is true in many areas of life, but particularly (and as we were discussing) in hockey. Numbers and performances tend to want to regress to their mean, and outliers will eventually be brought back down to earth. Such is the way of the universe. Or... something like that.

Over the past few days, we took a look at how the Over the past week we’ve taken a look at how the Minnesota Wild’s forwards, defensemen, and goaltenders fared over the past season, and now it’s time to take a look forward. What do these performances tell us about what’s to come? What’s been anomalous, what’s been about par for the course, and how might we expect things to stabilize?

We’re going to start in the crease and work our way outward, so let’s talk netminders!

All told, both Devan Dubnyk and Alex Stalock had down years last season. Dubnyk, who maintains a career save percentage of .916 and has posted save percentages of .936, .918, .923, and .918 (again) over his previous four seasons with the Wild, put up just a save percentage of just .913 last season.

Such was also the case with Stalock. He holds a career save percentage of .909 and posted save percentages of .932, .902, .884, .910 in his last four seasons in which he played 10 or more games. He finished last season with an .899 save percentage — not the lowest of his single season averages, but still well below the career average.

With this taken into consideration, we might expect that with both goaltenders posting season save percentages well below their career averages, they would be able to improve those next season, at least pulling them closer to that average. Regression to the mean, things of that nature.

But it’s also worth nothing that regression to the mean doesn’t necessarily account for the aging curve beginning to strike. And this becomes more (crucial) when looking at Dubnyk’s performances. He’s the unequivocal starter of this team, and has shouldered a heavy workload during his time with the team, and as such has high expectations placed on him to perform well. And while a .913 record is something of an outlier for him, the fact also remains that we’re looking at a 33-year-old goaltender whose numbers have been declining over the past three seasons. Fans may well still see a bit of a bounce back from him, but it’s equally likely that he’s doing what all aging netminders are wont to do — declining.

The same, really, could be said of Stalock as well. His career numbers have seen much larger swings from season to season, and given that history, an upward swing feels the more likely response. However, he’s also just turned 32, and as such is exiting his prime, and his ability to bounce back in a big way decreases with that. Given the number of games Dubnyk is tasked with starting, there isn’t quite as much riding on Stalock for bouncing back, relatively speaking. But we should also prepare ourselves to see the results of the aging curve hitting them both in the not-too-distant future.

While we might be able to expect a degree of regression on the goaltending front, curiously enough the same can’t really be said of the defense. But that said, this might not be the worst news.

We’ll concede that it can be difficult to measure defense, so we’ll be doing our best to approximate, by looking at how well the Wild have suppressed offense. Over the last three seasons, they’ve been pretty consistent in how they’ve been able to limit their opposition’s offensive attempts at 5-on-5:

So the Wild this past season were just about in line with how much they’ve allowed in the way of goals and chances against at even-strength. As a result, it’s hard to expect them to regress away from that next season, particularly because we haven’t seen any sort of overhaul of the coaching staff or defense corps. If anything, they might slide back to allowing a few more shot attempts, but overall we can reasonably expect there to be a good bit of consistency from this season to next, just as there has been over the past few in these metrics. Regression, you’ve met your match on this one.

Anyway, let’s wrap this up by talking about the forwards and the Wild’s efforts to generate offense because this is perhaps where things get the most complicated. The Wild actually took a step forward in some metrics (available on Natural Stat Trick) on their 5-on-5 totals from the 2017-18 season:

And we should be able to feel pretty good about that, right? The underlying process improved, and that’s good, right?

But the curious part is that the Wild took a step back in the piece that actually wins them the games — goals. During the 2017-18 season, they ranked 13th in the league in goals for at 5-on-5 with 160, but this past season they only registered 140, which bumped them all the way down to fourth fewest in the league.

And this feels right about in line with what we touched on in our evaluation of the forward corps — that every forward not named Zach Parise shot below the league average of nine percent. So the forward lines just about as a whole weren’t getting the results to match their efforts.

The good news? Shot-based metrics tend to be more predictive of future success than goal-based results. This means that if the Wild are able to carry over this emphasis on generating a surplus of shot attempts and a good amount of high-danger shots, things should start to break for them sooner or later.

Fans do still have to worry about that pesky aging curve with the forward corps as well, but those lines do figure to have enough shooting talent on the whole to allow for some degree of bounce back. Which, if we may say so, is sorely needed.