Advanced stats predict conference finals of 2015 Stanley Cup playoffs
By Ian Cooper, Phil Curry, IJay Palansky & Mikal Skuterud
To begin with, Tampa Bay bested Montreal in six games, exactly as we predicted. Meanwhile, the Rangers-Capitals series went down almost as though we scripted it, with the guys in blue neutering the Caps’ allegedly potent offense, particularly their power play (which went 1 for 15 in the series), and requiring seven low-scoring one-goal games to finally send Ovechkin and Co. packing after yet another second round disappointment.
If that wasn’t good enough, our two finalists (Tampa Bay and Chicago) are now among the final four.
In the West, the Blackhawks proved to be even better than we imagined, dispatching the Wild in four games. Unfortunately, the Ducks continue to confound us with their now 8-1 record in the playoffs while our model continues to treat them as a bit of a joke, giving them only a 6.5% chance of hoisting the Stanley Cup next month (more on that below).
Speaking of which, we now have updated Stanley Cup probabilities.
Although we’re still projecting a Tampa Bay vs. Chicago final with the Lightning winning in seven, the Bolts are facing their stiffest competition this round; whereas, we’ve got the Blackhawks heavily favored against the Ducks. It would be cheesy to change our prediction now since both of our original picks are still in it, but our model now has the Hawks as the overwhelming favorites with a 36.7% chance of winning it all vs. 31.0% for the Lightning.
So what’s our crystal ball saying about this round?
Blackhawks vs. Ducks prediction: Blackhawks in 6
This Western Conference matchup that many fans and commentators have been hoping for and drooling over since the start of the playoffs is finally here. The perennial powerhouse Blackhawks against the consummate under-performer Ducks.
The pundits say this series may be a clash of hockey titans on par with last year’s terrific Western Conference Finals, which the Blackhawks lost in overtime in Game 7 to the eventual Cup champs, the L.A. Kings.
Bad news if you’re one of those folks. Our model predicts a dud of a series, as Chicago has a 74.2% chance of steamrolling the Ducks.
Our prediction of a one-sided series is probably surprising to many. True, Anaheim led the Western Conference this season with 109 points—seven more than fourth place Chicago. But as we’ve explained, regular season point totals don’t tell the whole story; in fact, they can be downright misleading. Over the course of the regular season Chicago had the upper hand in almost every other meaningful stat. The Hawks’ score adjusted shot attempt differential of 54.4% was good for second overall, and dominated the Ducks’ 15th place 51.1%. The margin in scoring chances was closer, but still tipped in Chicago’s favor at 52.5% versus 52.2%. And the Hawks’ 83.4% penalty kill bests the Ducks’ 81.0%.
However, the Ducks have already burned us twice by beating the Jets and Flames after we picked them to lose both series. So we dug a little deeper to see whether there was something unusual about this season’s Ducks that might explain why they seem to be outperforming our model.
As it turns out, there might be. There are two sets of variables that lead the model to constantly pick against Anaheim. First, as indicated above, many of the Ducks’ regular season stats are pretty pedestrian.
And second, our model absolutely hammers the Ducks because of their absurd 33-1-7 record in one-goal games. Historically a team’s record in one-goal games is mostly luck, so the Ducks’ success in them leads our model to significantly discount their regular season record.
Now, it’s possible that the Ducks’ fantastic record in close contests isn’t just luck. But even that wouldn’t explain Anaheim’s apparent departure from its regular season mediocrity as reflected in other stats.
Here’s the funny thing about that: It seems the Ducks’ game changed during the last eight weeks of the season, very much for the better. In other words, it’s possible that the Ducks in the playoffs are a different team than they were throughout much of the regular season.
The graphs from war-on-ice.com show the 20-game trendlines for the Ducks and Hawks for a few important stats. The first shows a surge in scoring chance differential for Anaheim. After dropping to below 50% in late February—meaning the Ducks were getting out-chanced by their opponents —they jumped to nearly 55%. From mid-February through the end of the regular season the Ducks were fourth overall in this important metric according to the site’s stats from war-on-ice.com.
(Red line for Chicago, orange for Anaheim)
This improvement in scoring chance differential was driven primarily by a new-found defensive emphasis, as shown by a steep decline in the number of shot attempts against the Ducks since late February, reflected in the graph below.
From our model’s perspective, the confounding issue may be that even after their underlying fundamentals improved, the Ducks continued to play—and win—an awful lot of one goal games, going 7-1 in March and April. Teams that find themselves in lots of one-goal games are generally not strong teams. Strong teams tend to win a lot of games by two or more. In their first 62 games, it could be argued that the Ducks were not a strong team as none of the hallmarks were there—robust possession statistics, limited shot attempts against, and generating better quality shots than one’s opponents.
In the last quarter all of those hallmarks started showing up. Yet the Ducks still found themselves in lots of one-goal games—and they still won them. This suggests that maybe that there’s something about how the Ducks play that makes them different from most teams who chalk up a lot of wins in one-goal games, and this difference may have masked the their underlying improvement and caused our model to discount their regular season results by too much.
In the end, we’re sticking with the model and predicting Chicago in six. But we’re hedging our bets a little and saying that we won’t be surprised if the Ducks make this a more competitive series after all.
Lightning vs. Rangers prediction: Lightning in 7
Aside from the fact that our model has picked the Lightning, if the Rangers win we’ll have to endure a whole bunch of blather about defense and goaltending always trumping offensive firepower in the playoffs (pure myth) and the importance of grit, determination, experience and general playoff mojo.
If the Rangers lose, there will be the odd nod to the Presidents’ Trophy Curse, which is basically just made up. I think we can live with that.
There will be other meaningless storylines around this series.
For example, we’re certain to hear a lot about whether Martin St. Louis—who has been underestimated at every stage of his career—has one last thing to prove to Tampa Bay GM Steve Yzerman after last year’s Olympic snub and trade. As though St. Louis doesn’t always try to win hockey games and is saving some special gear for his old GM.
There are lots of ways to spin this series, and however it turns out there will be a pat explanation for why the result was inevitable.
But here’s the bottom line: the best team (in our view at least) won each division in the East. And although the Bolts made the formidable Rangers D look like Swiss cheese and Henrik Lundqvist look like a sieve over a three-game span at the end of November, this one is likely to be extremely close.
The Rangers had five more regular season points and a slightly better penalty kill (84.3% vs. 83.7%), but Tampa Bay was a much better possession team, posting a score adjusted shot attempt differential of 53.3% (7th overall) vs. 50.6% (18th) for the Rangers. So far in the playoffs the Lightning have been winning without winning the possession battle, and Henrik Lundqvist has been stellar for New York, posting a 5-on-5 save percentage of 0.943 in the first two rounds. Even if Tampa Bay does get its possession game back on track, goals still won’t come easily.
Our model gives Tampa Bay a 53.0% chance of advancing, but nobody should be surprised if the Rangers, who have played 12 one-goal games this playoffs, make them work for every win and keep the Lightning’s offense very much in check over a seven game grind.
Here are the updated picks from Ian’s four-year-old (who was a perfect 4-0 during the last round, by the way).
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 (@ian_doha), a lawyer, former player agent and Wharton Business School graduate; Dr. Phil Curry (@phil_doha), a professor of economics at the University of Waterloo; and IJay Palansky, a litigator 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
Dr. Mikal Skuterud (@mikalskuterud) is a professor of economics at the University of Waterloo.