- Predicting a Stanley Cup winner is no easy feat, but here's what the numbers say to the Department of Hockey Analytics.
After six months and 82 games, lucky fans in 16 cities get to buckle up for a grueling eight-week ride known as the NHL playoffs.
A lot has changed since last year. Canada traded its shutout from the postseason for 5 of 7 teams in this year’s festivities. The Los Angeles Kings, who have 2 of the past 5 Stanley Cups to their credit, continued to dominate in terms of raw possession stats, but finally dispelled any myths that they were in the midst of a dynasty by missing this year’s playoffs and firing the men who brought them championship glory in GM Dean Lombardi and Coach Darryl Sutter.
Meanwhile, the Detroit Red Wings finally succumbed to gravity, ending a 25 year playoff streak.
The one thing that remains the same is our collaboration with our friend and Waterloo professor of economics, Mikal Skuterud, who helps us develop our playoff prediction model.
For those who have followed our methodology from the past two years, we have once again vetted dozens of regular season variables and looked at every playoff series since 2008 (a total of 135 at this point).
A handful of statistics gave us the best model, and they are (in no particular order):
1. Total Points on the season
2. Points over the last 25 games
3. Penalty Kill Percentage
4. High Danger Scoring Chance Differential (HDSCD, from naturalstattrick.com)
5. ESVA Corsi (from puckon.net)
6. DOHA Luck Score (a measure we developed that takes a team’s win percentage in one goal games and shootouts, and subtracts the average of its win percentages in games decided by two goals and three or more goals; a higher DOHA Luck Score suggests unsustainably lucky results)
All of the predictions below are based on our model, not gut feelings or new information that we can’t quantify.
We also asked Ian’s oldest daughter, who is now 6, to make her picks. And since the age of majority for fashioning oneself a hockey expert is 4 years old, her younger sister is making her own picks this year. Break out your crayons and see how you stack up against them.
Now to the predictions…
Capitals (118 points) vs. Maple Leafs (95 points)
Prediction: Maple Leafs (50.5%)
We know–hometown fans. Completely delusional. Never gonna happen. We get it. But we were as surprised as anyone when our model spat out this result, and our computers don’t bleed blue and white. The Leafs limped into the playoffs by losing three of their four final games and appeared destined for a disastrous first round against the Capitals, who, unlike in their previous President’s Trophy winning campaigns, have looked disciplined, deep and defensively responsible. The Capitals topped the NHL in ESVA Corsi (54.9%), compared to Toronto’s 51.6%. But where Washington has enjoyed an advantage in puck possession, Toronto has dominated on shot quality, posting a HDSCD of +86 to Washington’s -64. Toronto also had a league-worst record in one-goal games (.378), which included 15 overtime/shootout losses and a 1-8 record in shootouts. Last time we checked there weren’t too many 3-on-3 overtimes or shootouts in the playoffs, so that performance says nothing about what will happen this week. It’s also worth noting that while our model prefers the Leafs in this matchup, it doesn’t view them as an unambiguously stronger team. For example, when we looked at second round matchups, the Capitals’ odds of beating the Canadiens or the Bruins in a Conference Final were slightly better than those of the Leafs. All of this suggests that this series should be closer than many people think and according to our model is essentially a coin toss.
Penguins (111 points) vs. Blue Jackets (108 points)
Prediction: Penguins (55.3%)
The defending Stanley Cup champs against a team that walked into last June’s draft lottery with a decent ticket for the Matthews/Laine sweepstakes and walked away with the rights to Pierre-Luc Dubois, who underwhelmed in his return to major junior and some are prematurely labeling a draft bust. What a difference one pick makes. Yet somehow the Blue Jackets shook off the potential “laughing stock” label and put together a terrific season in which forward Alexander Wennberg emerged as an elite talent, defenseman Zach Werenski quietly put together a campaign that would be Calder worthy in most years but was overshadowed by the Matthews/Laine storyline, and the rest of the lineup played great. Pittsburgh enjoys a slight advantage (+0.5%) on ESVA Corsi and is a far more dangerous team (+112 on HDSCD vs. +33). The Penguins also earned two more points in their final 25 games but have a far worse penalty kill, leaving the Blue Jackets with a +2.7% advantage. Of course what our model doesn’t fully appreciate is that Penguins defenseman Kris Letang is not going to play this postseason. His absence is partially reflected in the data (he missed half the season and most of the final 25 games), but add on the fact that Evgeni Malkin is nursing a sore shoulder (but should play in the first round), and you could be forgiven for concluding that this series might end differently and give Columbus fans a bit of a playoff run.
Canadiens (103 points) vs. Rangers (102 points)
Prediction: Canadiens (60.4%)
Those who follow our column will remember we’ve been consistently tough on the Habs. We predicted their collapse after a strong start last season and then explained what went wrong in February. With the return of coach Michel Therrien we didn’t expect much from them this season either, but then GM Marc Bergevin finally cut bait on Therrien, hired recently fired Bruins coach Claude Julien and things immediately began looking up in Montreal. The Canadiens still have weaknesses, but they dominate the Rangers handily on ESVA Corsi (+3.9%) and have a slight edge on HDSCD (+51 vs. +27) and on penalty killing (+1.3%). Rangers fans will dust off the old “King Henrik” routine, and they are surely right–Lundqvist is a great goalie. But you know who else is a great goalie? Carey Price. Advantage Montreal.
Senators (98 points) vs. Bruins (95 points)
Prediction: Bruins (74.7%)
Parity may rule in today’s NHL, but not every series is a coin toss. On paper the Bruins are simply a far better team than the Sens, dominating by an almost shocking 6.0% on ESVA Corsi. However, the Bruins don’t simply get more shot attempts than the Senators, they also are far more dangerous and enjoy a +108 to -68 advantage on HDSCD. When Boston winger Brad Marchand inevitably gets sent to the penalty box for something, the Bruins are also better positioned to handle things with their league-leading 85.7% penalty kill, whereas Ottawa struggled in that regard at 79.7% (22nd). The Bruins also earned three more points in their final 25 games, which suggests the team responded well after firing Julien. No offense to Julien, who is a terrific coach and deserves credit for getting things back on track in Montreal, but depending on who you ask, the B’s are either reaping the inevitable benefits of Julien’s system or actually playing better hockey under interim coach Bruce Cassidy. In either case, to the extent there are any easy outs in today’s NHL, the Senators should be one for them.
Blackhawks (109 points) vs. Predators (94 points)
Prediction: Predators (65.0%)
Though we’re sure some people don’t, it’s hard not to like the Blackhawks. This well-run team has become a fixture in the postseason, delivering an Original Six franchise half of its total Stanley Cup tally in the past seven seasons. So is this the first chapter of another epic playoff run? According to our model, not so much. The Blackhawks continue to dominate in terms of pure possession, enjoying a 1.1% advantage over the Preds on ESVA Corsi, but where they have succeeded on shot quantity, they have struggled mightily in the battle for quality shots, posting a -123 HDSCD vs. Nashville’s +36. The Predators have a middling penalty kill (80.9%), but Chicago has quietly gone from being a penalty killing machine to a bottom-dwelling 77.7% (24th). We were as surprised as anyone to find the Hawks rubbing elbows with the likes of the Sabres, Jets, Coyotes and Avalanche in this regard, but the numbers suggest something has been going horribly wrong when the Blackhawks are a man down. As we’ve written in the past, Nashville’s entire system seems to be built around using four elite defensemen to limit the quality of shots Pekka Rinne faces. Unfortunately for them, they went through a 31-game midseason stretch in which they suffered sequential losses to defencemen P.K. Subban, Ryan Ellis and Roman Josi, a fact that suggests their point total is understated. It’s worth noting that when they got past those injuries, they earned only two fewer points than Chicago in their last 25 games. Chicago certainly has a lot of talent and it's still young enough to prove our model wrong, but there are lots of signs to suggest that the team with the best record in the West has lots of vulnerabilities and will fall prey to its aptly named divisional rivals.
Wild (106 points) vs. Blues (99 points)
Prediction: Wild (71.7%)
There are a whole bunch of interesting storylines that will get bandied about in this series. The Wild had the second-best regular season record in the West, but will this be yet another postseason disappointment for coach Bruce Boudreau? Meanwhile, the Blues decided to speed along their transition from legendary coach Ken Hitchcock to ex-Minnesota bench boss Mike Yeo midseason. But there’s another storyline that won’t get mentioned in many circles and that’s the fact that toward the end of last season the Wild hired A.C. Thomas and Alexandra Mandrycky, who co-founded the now defunct war-on-ice. Many (including us) would count these two as among the smartest people in hockey, and while we have no inside knowledge of what actually goes on in the Minnesota front office, it appears they are making an impact.
You may wonder why a team that has embraced analytics sits 20th in the league in shot attempt percentage (a mere 49.4%), but when you look at ESVA Corsi, the Wild improve significantly to 51.6% (ninth) vs. the Blues’ 50.6% (14th). More interesting, the team that hired the people who began tracking high danger scoring chances led the league in HDSCD with a monstrous +188. The Blues managed to tread water in this regard (+1). We have some other reasons to be excited about the Wild. For example, early on this season, despite the punishing impact of age on pretty much every player, we labeled Eric Staal the most likely veteran player to see a rebound. Staal exceeded our 44 point projection by a mile, and teammate Jason Pominville, who we had as our fourth-best bounce-back player, also exceeded our 42 point projection. Meanwhile, Mikael Granlund, who we considered an offer sheet target before Minnesota signed him to a bridge deal two seasons ago, has come into his own, as have youngsters Nino Niederreiter, Charlie Coyle and Jason Zucker. Minnesota’s terrific season also was accomplished with a .475 record in one-goal games. When games are decided by three or more goals they boast a league-best record of .818. St. Louis does enjoy a +1.9% advantage on the penalty kill and earned 10 more points in their final 25 games, which suggests the midseason transition to Yeo helped. But by nearly every measure the Wild are fast, skilled, dangerous, deep, and likely to win this series.
Ducks (105 points) vs. Flames (94 points)
Prediction: Ducks (71.9%)
This series seems to have all the makings of a lopsided affair. The Flames appear to be the weaklings of this year’s playoffs. Like most teams coached by Randy Carlyle, the Ducks have been far from a puck possession juggernaut, but because the Flames were 16th-best in the league on ESVA Corsi, the Ducks actually enjoy a 0.3% advantage. The Ducks make up for their middling shot quantity on quality, with a solid HDSCD of +22 vs. Calgary’s -19. The Flames had a strong finish to the season, earning 35 points in their last 25 games, but guess what: so did the Ducks, who wound up with 37. Anaheim also enjoyed a +3.1% advantage on the penalty kill. The Flames were also extraordinarily lucky in one goal games, achieving a .690 win percentage. Fans certainly love watching those close games, but as we’ve said repeatedly, eventually that one-goal game magic runs out, usually in the playoffs.
Oilers (103 points) vs. Sharks (99 points)
Prediction: Sharks (51.3%)
Last year’s Western Conference champions limped into the playoffs with 25 points in their final 25 games, while the Oilers roared into a berth with 35. Meanwhile, both teams had a 80.7% penalty kill and boasted an impressive advantage on HDSCD (115 for San Jose vs. 72 for Edmonton). The Oilers were a decent 51.0% on ESVA Corsi (12th overall) but San Jose was better in this regard, posting a 6th best 52.2%. Where our model really differentiates these two teams is on our luck score, which punishes the Oilers for their .553 win percentage in one-goal games and rewards the Sharks for their .475 record. Of course, there’s always the possibility that Connor McDavid and Leon Draisaitl continue their upward trajectory into the playoffs and Edmonton just continues to get better while the Sharks finally succumb to the grind of playing more or less without a break for two straight years. The series looks to be fairly close to a coin toss, and promises to be fast-paced, exciting, fun, and worth staying up late for if you live on the East Coast.
And the Stanley Cup goes to…
If you didn’t get the strong hint before, we think the Wild have put together a tremendous season and are the standout team in the Western Conference. The only things that make us nervous about the 29.6% chance our model gives them of making it to the Final and the 19.2% chance it says they will be the star attractions of a victory parade in June are: (i) they stumbled down the stretch; (ii) they’re facing a very good coach in the first round (coincidentally one they fired last year) whose team rebounded immediately after he assumed the job; and (iii) those are unusually high odds in a league with so much parity. Nevertheless, we checked the numbers multiple times, so we’re sticking with that prediction.
The Eastern Conference, meanwhile, is a far messier affair in that, for the first time in a while, it appears to be stronger than the West. Our model nominally favors the Bruins, but that’s only due to the fact that the Leafs-Capitals series is expected to be so close. Once a winner emerges from that contest, the odds reset. If the Leafs win, Boston retains a slight edge. If it’s Washington, the Caps become the favorite in the East.
So if you’re picking a hockey pool, there are no easy answers in the East. You could be forgiven for picking Bruins, or you could flip a coin, decide the Caps will win the first round, and pick them to go the distance.
In either case, Minnesota wins.
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.