Analytics: NHL playoff probabilities at the quarter mark
We at the Department of Hockey Analytics love the holidays. Good food, family, large credit card bills - it’s our second favorite season. It’s also the time that our thoughts turn to our favorite season: the postseason.
It is around this time that various stats start to approach their peak predictive power. Once teams have reached the quarter mark of their schedule, we start to get a strong sense how they’re playing, and thus which have the best chance of making the playoffs. And so we find ourselves wondering, are the Blue Jackets for real? Can Detroit continue its incredible 25-season streak of making the postseason? Are there any teams currently on the outside that are poised to make a serious run? Why don’t we have any friends?
The answer to the last question may be fairly obvious, but for others, we turn to good old fashioned statistical analysis. When assessing a team’s playoff chances, what are the things that one should look for?
Unsurprisingly, the stats confirm that the teams that make the playoffs are the ones that are best at controlling the play, generating good scoring chances, and limiting scoring chances against.
The best measure that we have for how well a team controls the play is Corsi. Shot attempts appear to be correlated with time of puck possession, but perhaps more importantly, shot attempts are an indication of meaningful possession since, by definition, the possession resulted in a shot attempt. However, since teams change the way they play depending on whether they are leading or trailing late in the game, or if they are home or away, the current state-of-the-art is to make adjustments to the stat. For these factors using a Corsi variant called Event, Score and Venue Adjusted Corsi (ESVA Corsi) gives a better picture of how a team is doing. For those interested in finding out more, PuckOn.Net is an excellent resource.
When it comes to examining the quality of scoring chances, there are several options. There are a number of people who have developed their own measure of expected goals, which weights each shot with the likelihood that it might go in. Brian Macdonald, who is currently the Director of Hockey Analytics for the Florida Panthers, was one of the first to go public with a method for calculating this stat. Currently, measures of expected goals can be found on Corsica.Hockey, run by Emmanuel Perry, and on Don’t Tell Me About Heart, run by Dawson Sprigings (and probably elsewhere as well—we apologize if we missed your site). Corsica also has expected shooting and save percentages, but the measures that we found to have the most predictive value were just regular scoring chances for and against, as found on Corsica.
Finally, one cannot ignore the standings. Not only do the standings contain information missed by the above stats, the points banked (or not earned) go a long way to determining whether a team makes the playoffs. A lucky start that eventually stalls may not portend a long playoff run, but it may be good enough to allow a team to limp into the postseason.
So, to summarize, we looked at many variables, but the model that was best at predicting which teams would make the playoffs based on past years’ results used 4: team points, ESVA Corsi, scoring chances for and scoring chances against for each team. All stats were taken after each team’s 25th game since 2007-2008 to 2015-16 (excluding the lockout shortened season of 2012-13). Based on those past seasons, we developed a model to predict the likelihood of making playoffs. The model gives us the following probabilities of making the playoffs this season:
Eastern Conference
| Team | Pts. after 25 games | Prob. of Playoffs | Team | Pts. after 25 games | Prob. of Playoffs |
| CBJ | 34 | 77.2% | MTL | 36 | 71.6% |
| NYR | 33 | 73.9% | OTT | 32 | 42.7% |
| PIT | 33 | 72.3% | BOS | 29 | 69.8% |
| WSH | 33 | 72.5% | TB | 27 | 44.3% |
| NJ | 30 | 30.6% | FLA | 26 | 45.6% |
| PHI | 27 | 47.0% | DET | 25 | 10.5% |
| CAR | 25 | 43.2% | TOR | 25 | 62.5% |
| NYI | 25 | 7.2% | BUF | 24 | 29.4% |
Western Conference
| Team | Pts. after 25 games | Prob. of Playoffs | Team | Pts. after 25 games | Prob. of Playoffs |
| CHI | 35 | 92.7% | SJ | 31 | 93.5% |
| STL | 32 | 85.4% | ANA | 29 | 70.4% |
| MIN | 30 | 80.6% | EDM | 28 | 84.2% |
| NSH | 28 | 85.6% | LA | 28 | 86.0% |
| DAL | 24 | 17.3% | VAN | 24 | 9.2% |
| WPG | 24 | 21.2% | CGY | 22 | 6.4% |
| COL | 21 | 3.1% | ARI | 20 | 0.1% |
First, it is worth mentioning that after years of dominance, the Western Conference is clearly weaker than the East. Moreover, there seem to be only eight teams that have a legitimate shot at the playoffs, meaning that we shouldn’t expect much drama down the stretch in the West. The Winnipeg Jets and Dallas Stars could pull things together to make things interesting, and it is worth remembering that these predictions are based on the first 25 games, so they don’t take into consideration the Calgary Flames’ recent hot streak that saw them get all 12 possible points in their six games since. The Flames have certainly increased their chances, especially since over this winning streak their underlying stats were much improved. Their ESVA Corsi was up to 51.6% over those 6 games, as opposed to 48.2% in the previous 25, and they also managed to get on the positive side of scoring chances generated (60 for to 57 against), while they had been buried in that regard previously (190 for to 246 against). If they can maintain this improvement, they may be the team with the best chances of forcing their way into the playoff picture.
In the East things are much more interesting. This year’s surprise story, the Blue Jackets, appear to be for real, but the conference is so deep that no team has as strong a chance as even the seventh best team in the West. Of the teams off to strong starts, the Ottawa Senators and New Jersey Devils look most likely to be in for a collapse, although Corey Schneider certainly has the potential to carry the Devils all on his own, even though he has been struggling so far.
Somewhat surprisingly, our model tells us that the Toronto Maple Leafs are the strongest contender to fight their way into the playoff picture. Jeff Veillette, editor at The Leafs Nation, has posited that #TheLeafsAreActuallyGood, and the data seem to indicate that he certainly has a chance of attending Game 3 of the first round of the playoffs this year. Like Calgary in the West, however, The Philadelphia Flyers have done the most to increase their playoff chances in the games after the 25-game mark. As of this writing, they are on a nine-game winning streak, with six of those wins not included in this analysis. In Philly’s case, however, their underlying stats in their winning streak are pretty much identical to the previous 25 games, so the probability given above may be closer to accurate than Calgary’s.
Finally, Detroit fans might want to prepare to experience something that some have never seen before: a season with no playoff games. Indeed, they along with the New York Islanders are the two weakest teams in the East by a fair margin.
The Department of Hockey Analytics employs advanced statistical methods and innovative approaches to better understand the game of hockey. Its three founders are Ian Cooper, a lawyer, former player agent and Wharton Business School graduate; Dr. Phil Curry, a professor of economics at the University of Waterloo; and IJay Palansky, a partner at the law firm of Armstrong Teasdale, former high-stakes professional poker player, and Harvard Law School graduate. Please visit us online at www.depthockeyanalytics.com
