With the NHL All-Star break officially here, now is a good time to pull back and take a wide angle view of the league.
To do that, let’s look at a few plots based on a power ranking of all 31 teams using a Bradley-Terry model. For more information on how the model works, refer to this article. But in simple terms, we’re employing a modeling framework that uses individual game outcomes to determine which teams are the best. The model accounts for venue (home/away) and margin of victory. This approach also has the benefit of including a built-in quality of competition so that playing well against a good team is rewarded more than playing well against a bad team.
To start, here’s how the teams rank using both individual game expected goal and actual differential. All data for this article is via Evolving Hockey.
For Lightning fans, this should be a nice chart to see. The Bolts are the best team in the league by the model’s calculation based on their game by game expected goal differentials. And in terms of actual goal differential, they rank third behind the Bruins and the Blues.
The teams with big gaps between their rankings in terms of expected goals and actual goals are also interesting. The Kings are a team that appears to control play but can’t translate that into results while the Jets seem to be stealing lots of games they have no business winning.
One other thing that stands out is that four of the bottom eight teams are from the Pacific division, including the Canucks who currently sit in first place. Wow, is that a bad division.
The next plot shows the distribution of individual game performance both in terms of expected goal and goal differential. The teams are ordered by their power ranking from the above plot.
The first thing to notice here is the general observation of how much more variable goal outcomes are than expected goal outcomes. This is a helpful reminder of how much of what happens in a hockey game is dependent on variance.
The Wild have an extremely pleasing normality to the shape of their expected goal distribution. The Avalanche have played lots of very close games in terms of shot metrics. The big disparity between the Kings quality of play and their results is visible again here.
The next plot shows the distribution of the strength of schedule for all teams.
Because all 31 NHL teams play each other during the season, we don’t see much variation this far into the campaign. The teams here are ordered from strongest to weakest average quality of competition in terms of expected goals and the vertical line marks that average for each team. But the difference between the top and bottom teams is minimal and by the end of the season, will be negligible. Keep in mind this measure of strength of schedule doesn’t include whether the team faced an opponent at home or away and it doesn’t include any measure of rest or travel so I still have work to do to improve this.
And finally, here is how each team performs against different levels of competition. Again, we’re using the expected goal ability estimates to define team quality. A fit line that slopes downward means a team beats up on bad competition and/or struggles with tough competition. A flat line means they play the same against all levels of competition.
No particular shape of line is good or bad in this case. Blowing out bad teams isn’t a bad thing and neither is playing consistently regardless of competition. This view is purely a descriptive way of seeing how each team has performed against different levels of competition.
From this plot, it appears the Leafs and the Golden Knights have the biggest gaps between how they play against good teams compared to how they play against bad teams. The Lightning have a relatively flat line meaning they’ve played to a similar level against varying levels of competition.
By now, you might be wondering if this article was just an excuse for me to post some charts I like on a Friday. And the answer to that question is yes. Yes, it was. So with that accomplished, let’s all enjoy the All Star festivities. Or not. It’s really up to you.
