Evaluating October hockey is tricky
Published
The trickiest part of evaluating hockey in October is differentiating what’s real and predictive of future performance, and what’s us reading too much into small samples of data.
Consider the Pittsburgh Penguins and Toronto Maple Leafs as a quick example, two teams that have played the probably tanking Arizona Coyotes. Pittsburgh won in a four-goal blowout. Toronto suffered a two-goal loss.
We know that’s an expected win by Pittsburgh and a disappointing loss by Toronto – if you listened to Maple Leafs’ head coach Sheldon Keefe, disappointing may be an understatement. And yet we think of Pittsburgh and Toronto as comparably skilled teams, with oddsmakers willing to give an edge to the Maple Leafs.
What do we do with these results, and more interestingly, how do we go about evaluating teams so early in the regular season? Which teams offer opportunities to develop confidence in the results, and which teams may be getting misleading results right now?
This isn’t an exact science, but one of the ways I like to measure team quality – early, and often – is to study a team’s performance by isolating for one major variable. We know hockey is a lower-scoring sport, and with that comes more noise and volatility. Controlling for that volatility, especially just 3 per cent of the way into the regular season, is essential.
The below table shows each team’s rank relative to their peers by goal differential. The goal differential is four-pronged: we measure actual goal differential, goal differential while controlling for goaltending performance, goal differential while controlling for shooting performance or conversion rates, and expected goal differential, controlling for both goaltending and shooting.
As you would expect, adjustments aren’t significant for most teams. But a few teams have notable outlier performances, and I’ve highlighted them accordingly:
Teams like the Penguins (2-0-1) are off to dominating starts, and incidentally, they look great across the board. Controlling for their goaltending play by using expected goals against, or controlling for their offensive play by using expected goals for, offers little change in our understanding of the team to date. Why? The Penguins are playing a significant amount of minutes in the offensive zone, generating heavy shooting volumes while holding their opponents to little, and are driving favourable goal differentials as a result.
Compare Pittsburgh to other teams with little change in their rankings around the league – the Washington Capitals (2-2-0) have played mediocre hockey so far, and that’s true no matter which way you look at it. The Anaheim Ducks (1-2-0) have been abysmal, also true from every possible lens.
Now, let’s look at the three teams where our understanding may be more ambiguous:
Minnesota Wild (0-3-0)
Minnesota has been outscored 20 to 12 (-8) in their opening three games, and head coach Dean Evason is frustrated. Prior to the regular season, it was anticipated that Minnesota would ice a quality team more likely than not to qualify for the postseason. But that’s not happening with goaltending like this. The Wild have a 34-shot advantage over their opponents and should be near break-even in expected goals, but Filip Gustavsson (.860) and Marc-Andre Fleury (.776 stop rate) have been abysmal. There’s no reason to be concerned with the Wild’s skater group quite yet, but one thing is true for Minnesota and any other team in the league: If you can’t stop the puck, you aren’t playing meaningful hockey come April.
New York Islanders (1-1-0)
The Islanders have just a pair of games under their belt, but they are the inverse version of the Wild. Ilya Sorokin has been sensational to start the year, stopping 94 per cent of shots and erasing more than four goals versus expectations against the Florida Panthers and Anaheim Ducks. What’s remarkable is despite Sorokin’s play, the Islanders only have two of a possible four points. That’s because this is the type of duress they are putting Sorokin under (via HockeyViz).
New Jersey Devils (0-2-0)
There are serious expectations on the Devils to deliver a better product on ice this year, but so far, Lindy Ruff’s team has taken a couple of steps backwards. The Devils are actually carrying a positive expected goal differential (+2) over the first two games, but have been outscored 10-4 (-6) in a pair of games against the Philadelphia Flyers and Detroit Red Wings. No team in the league can string together wins while scoring on just five per cent of their shots, but a primary concern with this Devils’ lineup was if they had enough top-end shooting talent in their forward pool – the type of talent that can take advantage of territorial advantages and bigger volumes of scoring chances. Of their four goals, two (Damon Severson and Dougie Hamilton) have come from defencemen. Not the start they wanted.
Data via Natural Stat Trick, Evolving Hockey, NHL.com, HockeyViz