Happy Max Corsi Productivity Day! We’ve reached the point in the season where Corsi best predicts future winning percentage. There’s plenty of more advanced ways to better predict how the rest of the season will go, but Corsi offers a simple baseline in a way that helps explain why it is so important. I’ll first explain what that means and why it matters, then take a look at how we can use it to predict basic shifts in the standings for the rest of the NHL season.
Corsi and Predictability
People who study hockey from a statistical perspective talk a lot about Corsi (shot attempts). Why worry about shot attempts since winning is what matters, and winning is decided by goals? We use Corsi because goals can be streaky. Sometimes a player has the golden touch, and other times a goalie is dominant. Moreover, goals are so rare that it can take a long time for a hot or cold streak to be balanced out by more representative results.
That’s where Corsi comes in. Corsi is based on all shot attempts, so we get a large sample of them much sooner than we get one by measuring goals. The large sample then makes it easier to predict future results than we could with just goals or wins, and that’s what we’ll do here. (For more on the importance of shot volume and shot quality, I’d recommend this piece by Eric Tulsky and the included links.)
We’re choosing this point in the season because the data tells us that the optimal time to make these predictions is 20-25 games into the season. Do it earlier and the sample size would be too small for optimal results. Do it any later and there are too few games left to predict. This latter part is less intuitive, but imagine trying how hard it would be to predict the final game of the season based on the first 81.
This isn’t to say that predictions today will be dramatically better than predictions last week or next, but this is the time when we’ll have maximum (though still limited) accuracy.
Today’s Corsi, Tomorrow’s Wins
What changes in the standings seem likely based on shot differential? First, here’s the current standings as of Sunday, ranked by the percentage of possible points that each team has gained:
|Eastern Conference||Western Conference|
|Rank||Team||Point %||Rank||Team||Point %|
Now, here’s what the standings would look like if hockey was played in the Corsi Hockey League, i.e., with all the teams ranked by their 5v5 Score-Adjusted Corsi For % from corsica.hockey. I’ve also added a column indicating how they’ve changed in these conference rankings compared to the real ones. This isn’t everything that stats can show. It’s just a simple first-look based on Corsi alone.
|Eastern Conference||Western Conference|
A few teams stand out here:
- The teams with the most dramatic swing are the Rangers and Hurricanes. New York is 3rd in the East by points and 14th by Corsi%, while the Hurricanes are the reverse. Goaltending differences certainly explain a portion of this, but we can predict that both of these teams will move closer to the middle of the pack
- Several teams in the east besides Carolina are underperforming their shot differential: Boston, Washington, Florida, Philadelphia, and Buffalo (albeit just to mediocre in this last case). All of them have reasons for optimism, especially if their goaltending holds
- On the other hand, Ottawa, Montreal, and New Jersey may struggle to maintain the pace they’ve had so far
- In the west, the standings match shot differential much more closely than the east. I’d expect these standings to stay more static, though we could see both Los Angeles and Nashville rise while Chicago and Anaheim fall
- Detroit, Phoenix, Colorado, and the Islanders are bad and should feel bad
Obviously, none of this is a guarantee. As Hockey Graphs Chief Computer Guy Garret Hohl says, we’re talking about probabilities, not destinies. The prediction I can make with the most certainty is at least one of the above comments will be wrong. But, on the whole, this will likely be more accurate than expecting the standings to stay the way they are.
This is also a super simple way of assessing a team’s performance. There are a lot of more complex details that add value. For example, expected goal models like those from DTMAboutHeart and Manny Perry add a shot quality element to Corsi’s shot volume.
You can also break down team results into its components. For example, Carolyn Wilke creates handy charts which show each team’s scoring chances and PDO. I also like Sean Tierney’s Team At A Glance visuals. Using all of these together will give you a fuller picture of what to expect for the rest of the season.