NHL playoffs: Analytics predict East's second round
Round 1 in the Eastern Conference went almost perfectly according to plan. As our model predicted, the Penguins, Capitals and Lightning romped, and the near coin flip we predicted the Islanders-Panthers series to be ended up being decided by back-to-back double overtime games.
So the chaff hasn’t just been separated, it’s been unceremoniously slam dunked in the dumpster. The Red Wings, Flyers, and Rangers never really had a chance. No more soft spots means no more easy picks. But it also means two fantastic matchups, including a Capitals-Penguins duel that we think may well decide this year’s Stanley Cup champ.
Capitals vs. Penguins
Prediction: Penguins (54.7%)
Admit it, even if you’re a Rangers or Flyers fan, this is the series you’ve been drooling over for weeks. Crosby versus Ovechkin. Presidents' Trophy winner versus second-half powerhouse. Supporting casts that read like All-Star line-ups, with Malkin, Letang, Kessel, and Hornqvist for the Pens, and Backstrom, Kuznetsov, Carlson, Oshie and Holtby for the Capitals.
Four months ago the Capitals were on a record-setting pace and looked unstoppable, while murmurs were growing louder that Sidney Crosby had not only fallen from his perch as the world’s best player, but that his play had dropped off so much that he probably wasn’t even good enough to make Team Canada’s World Championship team. There was serious doubt about whether the Pens would make the playoffs at all, let alone contend for the Cup.
But then the Pens hired Mike Sullivan as their new head coach. Crosby returned to his usual, dominating self. And before you knew it the Penguins were steamrolling teams on their way to a fourth place overall finish despite their crummy start and despite missing Evgeni Malkin for most of the last part of the regular schedule.
The stats our model found to be predictive of playoff success appear to be pretty close for these two teams. Las Vegas pegs the Capitals as a solid favorite, but our model points in the other direction, and gives the Penguins a 54.7% likelihood of coming out on top. If anything, there’s reason to think that our model may not give the Pens as much credit as they’re due because they have probably been the best team in the league during the last few months, but the inputs to our model are mostly teams’ full-season stats.
If we focus on just a few key stats since January 1, we see that the Capitals were still pretty good, but the Penguins were great. As the tables below show, before January 1, the Penguins had a negative goal differential (-3), and were 22nd in the league in ESVA Fenwick and 13th in high danger scoring chance differential (HDSC +/-). Since January 1, they were second in each of those categories, including boasting a massive HDSC differential almost seven times that of the Capitals.
These underlying numbers translated into points, with the Penguins earning five more than the Caps over the last 10 games of the regular season, a stat that was clearly a contributing factor in our model's prediction. Whether or not you believe momentum is a real thing, it’s clear the Penguins under Mike Sullivan are playing their best hockey in years. This series will be a hard fought battle between the two most skilled teams in the league. Our model picks the Penguins to survive.
Lightning vs. Islanders
Prediction: Lightning 57.2%
Before the playoffs started things looked a little iffy for the banged-up Lightning. But with the re-addition of a rejuvenated Jonathan Drouin and big numbers from Nikita Kucherov and Tyler Johnson, the Lightning’s 4-1 series win over Detroit seemed to validate our model’s 70.2% confidence that they’d advance against the Red Wings.
On the other side of this bracket, the Islanders rode the heroics of John Tavares and 30-year-old journeyman goaltender Thomas Greiss (.944 save percentage in the first round vs. his .917 career average) to pull out a win in a series in which they were generally outplayed by the Panthers.
As for this series, the Isles compiled more points during the regular season (100 to Tampa Bay's 97) and actually had better penalty kill, power play, and xSv% numbers. But the Lightning dominated in most other categories, in particular penalty differential (+25) and high danger scoring chances, where the Islanders for the season were a dismal -31, compared to the Lightning’s +46.
Our model and the eye test would seem to agree that the Lightning should take the series handily, but if we dig a little deeper the outcome may not be so certain. Our model doesn’t take injuries into account and the Lightning are still missing key players Steven Stamkos (blood clots) and Anton Stralman (broken leg). And although the Lightning won their first round series in just five games, the advantage we expected to see in possession and shot quality didn’t materialize. Tampa was actually -13 in high danger scoring chances and lost the possession battle as well, with a score adjusted Fenwick For % of 49.8.
Luckily for the Bolts, the Islanders aren’t a much stronger team than the Red Wings and were probably lucky to get out of the first round. For New York to win this series, Thomas Greiss will have to play like his injured battery-mate Jaroslav Halak did when he was with Montreal and stole back-to-back series from heavily favored Washington and Pittsburgh in 2010. Although no model can accurately predict that sort of individual performance, we wouldn’t bet on Greiss bailing out the Islanders again in Round 2. But expect Tampa Bay to reach the Eastern Conference Finals for the second straight year.
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 (@ijay_doha), 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