Who advanced analytics (fancystats) predict will win the 2015 Stanley Cup
By Ian Cooper, Phil Curry, IJay Palansky and Mikal Skuterud
A hockey agent, a former professional poker player, and an economist walk into a bar ... No, this isn’t the start of a bad joke, it’s how we formed the Department of Hockey Analytics—a group dedicated to the belief that creative analysis of statistics can lead to a much better understanding of the game of hockey.
So here we are—and yes, the DoHA really does consist of a former player agent (Ian Cooper), a former professional poker player (IJay Palansky) and an economics professor (Phil Curry)—to take on SI’s hockey heavyweights in a Stanley Cup playoffs prognostication altercation.
For good measure, we’ve added a second economist, Phil’s friend and colleague Mikal Skuterud, to our posse.
During last year’s playoffs, when the Penguins were up 3–1 in their second-round series against the Rangers, everyone was talking about how brilliant Pittsburgh coach Dan Bylsma was for creating a superline of Sidney Crosby and Evgeni Malkin rather than limiting that dynamic duo to power plays. But we at the DoHA said, “Hold the phone: Based on our numbers, putting Crosby and Malkin together at even strength actually reduces Pittsburgh’s overall offense by killing the second line and beyond.”
What happened next? New York came back to win the series in seven games.
Fast forward to this year’s March 2 trade deadline, when the Pens had a solid chance of winning the Metropolitan Division. Again we sounded the alarm, saying that the team was in desperate need of scoring depth and urging Pittsburgh GM Jim Rutherford to trade for disgruntled Maple Leafs winger Phil Kessel. That didn’t happen, and ever since, the Penguins—the team with the best one-two punch in hockey—have struggled to score goals.
This season the DoHA not only predicted Toronto’s monumental collapse (OK, maybe that one wasn’t particularly tough), but also explained why hitting matters more than people think, why great offense may be more likely to lead to a championship than great defense, why the Avalanche went from the league’s No. 3 team last season to a bottom-feeder this year and why a team’s performance in its last 10 games of the season matters to its first round playoff prospects.
Now, before we get to our predictions, we need to get a couple of things out of the way.
First, predicting sports matchups enjoys a big advantage over other types of predictions: Even if you know nothing, odds are that you’re going to be right half the time. But being right 100% of the time is virtually impossible. To prove the point, we’ve asked Ian’s four-year-old daughter, who knows nothing about hockey except that watching it allows her to stay up past her bedtime, to make her own picks (you can see how she does on our blog).
Based on our math, we think that in an average year the DoHA will pick about 75% of the playoff matchups correctly, which means that we will be right about 11 series out of 15.
So how did we arrive at this year’s picks?
We looked at every playoff series since 2008 (a total of 105) to see whether there were things in the past that might help us predict the future. We examined traditional measures, including home ice advantage, regular season points, record over the last 10 regular season games, head-to-head record, power play percentage, penalty kill percentage and save percentage. We also took into account some newfangled stats, including Corsi For, Score-Adjusted Corsi, Shots-For percentage, PDO—the sum of a team’s shooting percentage plus its save percentage—and a measure of luck that we created ourselves based on a team’s record in games decided by one, two and three or more goals.
Look at the chart below and you’ll see what those different stats would have gotten you in the past in terms of correct predictions. If you had just picked the teams with the most regular-season points, for example, you’d have only gotten 54.3% of the playoff series since 2008 right—you could probably have done about as well by flipping a coin.
Comparison of Playoff Prediction Models
* Where the teams tied by a measure, the series is treated as a coin toss, meaning for every two ties, there would be one correct result predicted.
** The team with the better PP%, higher PDO or higher DOHA Team Luck Score actually lost more often.
Make sure your browser is able to load all scripts, and mouse over the bars in the chart to see individual figures.
If you think that power plays matter, you’re completely wrong. The teams with better regular-season power plays usually lost their series. Good penalty killing, on the other hand, matters a lot.
To get the best predictions, we combined multiple variables. If you’re into math, go to www.depthockeyanalytics.com to see exactly what we did.
Now to the fun part: what’s going to happen in the first round and beyond.
Prediction: Jets in 6
This series is a prime example of how looks can be deceiving. According to our model, Anaheim’s luck is about to run out. The Ducks’ No. 1 seeding is primarily the result of the team’s amazing 33-1-7 record in one-goal games. Many will say that this shows that Anaheim knows how to win, but analytics tells us that a team’s records in one-goal games is more a function of luck than anything else. Last year’s three best teams—the Kings, the Blackhawks, and the Bruins—were all below .500 in one-goal games. We suspected the Ducks might be headed for an early playoff exit, and our model confirmed our suspicions, picking upstart Winnipeg to advance.
Prediction: Blues in 7
It’s really too bad that these two teams have to meet in the first round, as both are legitimate Cup contenders. In fact, in the Western Conference, our model favors either of these teams over everyone but Chicago and, surprisingly, the Jets. Our model predicts that St. Louis will just barely scrape by into round 2.
Predators vs. Blackhawks
Prediction: Blackhawks in 7
Analytics show us that a team’s performance over the last ten games of the regular season is predictive of the outcome of its first-round playoff series, which is bad news for Nashville, which limped to the finish line (4-4-2) after a fantastic first three-quarters of the season. Luckily for the Predators, Chicago was even worse over its last 10 games (4-6-0). But the Blackhawks have better possession numbers, a better penalty kill, and are much less reliant on winning one-goal games. Our model gives them a 60% chance of beating Nashville and moving on. In fact (spoiler alert) our model picks Chicago to get all the way through to the finals. Oh, and Patrick Kane’s return to the lineup won’t hurt either.
Prediction: Canucks in 7
In a year full of surprises, one of the biggest is that these two teams are in the playoffs at all. Mind you, we sung the praises of Radim Vrbata as a hidden gem among available free agents last summer, and he has certainly delivered for Vancouver. While both teams deserve credit for making it through, the fact remains that they are the weaklings of the Western Conference. Nevertheless, SI has told that us we need to pick somebody to win this series. Our model says that it will be almost as close as the Blues–Wild series, and that the Canucks will squeak through to round 2, where they’ll either get smashed by the Jets or have a competitive series against Anaheim.
Rangers vs. Penguins
Prediction: Rangers in 7
Unless you’re a Flyers fan, it’s impossible not to feel sorry for Pittsburgh, which played great hockey most of the season but was decimated by injuries. The offensively minded Penguins have a surprisingly good penalty kill (84.8%) and have improved their possession numbers under new coach Mike Johnston. Unfortunately for them, their offense went off a cliff shortly after they ignored our advice and declined to trade for Phil Kessel at the deadline. New Yor, meanwhile, went from a possession juggernaut last season to being below average. If Pittsburgh had enough quality blueliners to suit up, this one might tilt its way. Regardless, the series should still be close.
Prediction: Senators in 7
Assuming the Hamburglar doesn’t suddenly turn into Mayor McCheese—and backup Craig Anderson isn’t pressed into service—it’s likely that this series will be a solid matchup between two great goaltenders. Ottawa has roared into the playoffs on the strong play of goalie Andrew Hammond, but it’s arguable that the Senators were a better team to begin with than the one that required a mid-season coaching change. Ottawa may be the better possession team, but for the past two years Montreal has shown that when you have Carey Price between the pipes, you can afford to get outshot. This series is unlikely to be a shootout, and in the end our model gives the Senators a 54% chance of pulling off an upset.
Prediction: Lightning in 6
There will be all kinds of narratives tossed around for this series: Steve Yzerman and Valtteri Filppula against their old mates in Detroit; youth vs. experience (coaches included); offense vs. defense. But here’s the bottom line: Tampa Bay is a really good team that trounced its opponents (.708 Win% in games decided by 3 or more goals, but only .528 in one-goal games), put up terrific possession numbers and had a solid penalty kill (83.7% to the Red Wings’ middling 80.9%). Detroit is not the Red Wings of old, and we give them a 64% chance of losing in the first round.
Prediction: Capitals in 6
Say what you will about Washington and its history of coming up short, but the Capitals made a number of offseason moves that have paid off. Like the Lightning, Washington didn’t ride a wave of coin tosses either, posting just a .457 winning percentage in one-goal games. Meanwhile, the Caps’ possession numbers went from appalling to pretty good. If you’re looking at regular-season points alone—both teams had 101—this one is literally a coin toss. But, like the Ducks, New York won a lot of one-goal games, and Lady Luck may already be turning against the Islanders given their tepid finish. We love a feel good story as much as the next guy—and certainly an epic playoff run that brings back memories of the great teams of the 1980s would be that. But with only a 39% chance that New York will take the series, it looks like Washington will spoil Long Island’s swan song.
And the grand prize winner is…
Picking a Stanley Cup champion on day one of the playoffs is a bit of a sucker bet. But if you have to pick one, the team with the best chance, according to our model (a walloping 14%!), is the Lightning, who we have edging out the Blackhawks in a seven-game series.
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.