NHL playoffs: Analytics predict West's second round
As we noted a couple of days ago, Round 1 in the Eastern Conference felt almost perfectly scripted. The West not so much. We knew the West was brutally deep, but we’d be lying if we said we thought the carnage would happen so soon.
The Anaheim Ducks, who we had going to the Stanley Cup Final, didn’t look particularly Mighty as they conceded the first two games to the Predators at home, woke up for a bit, and then managed a grand total of two goals when presented with two games in which to knock out the Nashville Predators. Highlights included a zero-goal stretch of seven games by Corey Perry (enough blaming the guy—cold streaks happen) and a bad penalty by Ryan Getzlaf in the dying minutes of Game 7, followed by a game of keep away as the Predators burned more time on a delayed penalty.
This despite being 58.7% favorites in our view.
It will be an interesting off-season for a team that boasts elite talent and depth (and until a few hours ago a very good coach) but has no hardware to show for it.
Meanwhile the Los Angeles Kings, who have been darlings among the analytics crowd for several years running, were sent packing by the Sharks. The surprise isn’t that San Jose overcame a 53.2% likelihood of losing the series—it’s that the Sharks made the Kings look like they didn’t belong in the postseason in the first place. With L.A.’s old backup goalie to rub some salt in the wound.
The Central Division contests played out as expected, with both Dallas and St. Louis advancing.
So what next?
Stars vs. Blues
Prediction: Stars (51.0%)
The hockey world won’t stop talking about the Capitals-Penguins series, and Game 1 surely delivered on its promise.
But all signs point to this series being a gem as well, and it would be a shame if fans overlooked it.
St. Louis and Dallas are nearly identical by a number of measures. Both had excellent regular seasons, finishing only two points apart, and both finished strong, each earning 16 points in their last 10 games. What you might not have noticed is that the Blues were more successful in the second half, earning eight points more overall.
Special teams are likely to be a storyline here. Dallas has a slight advantage on the power play (+0.6%), but St. Louis does a much better job of killing penalties (+2.8%). Maintaining a strong penalty kill will be particularly important for the Blues because while Dallas was roughly equal on penalties drawn vs. taken, St. Louis was -42.
Both teams had good possession numbers (ESVA Fenwick% of 52.8% for Dallas vs. 52.6% for St. Louis), but despite boasting one of the great burgeoning talents in the game in Vladimir Tarasenko, St. Louis had a high danger scoring chance differential of -66 vs. Dallas’s +38.
Two variables that really differentiate these teams are the DOHA Luck Score, which badly punishes the Blues for having a 0.658 win percentage in one-goal games but not faring nearly as well in games decided by a wider margin, and xSV%, where the Stars’ goaltending deficiencies were on full display in the final game of their series against the Wild, and the Blues enjoy a 2.0% advantage.
In his postgame interview Dallas keeper Kari Lehtonen seemed surprisingly unfazed by his implosion in the final game of the Minnesota series (four goals-against in the third period), perhaps because his mediocre first round performance (.911 save percentage) actually significantly elevated his career playoff average to a staggeringly bad .885.
The Stars seem to be performing an experiment on whether goaltending matters as much as people think it does. One thing is certain: They will need to score a lot of goals if they are to keep winning.
Our model doesn’t account for the fact that Tyler Seguin (Achilles injury) will miss the beginning of this series and his return remains uncertain.
In the end our model gives Dallas the slightest edge, but with an asterisk around Seguin that could tilt the actual odds in St. Louis’s favor.
Predators vs. Sharks
Prediction: Sharks (52.1%)
So here we are with two remaining teams in the Pacific bracket, both of whom we thought would be more likely on the putting green than the ice at this point. There will be no shortage of narratives that predict with equal measures of certainty either (a) yet another playoff disappointment for Joe Thornton & Co.; or (b) that the Sharks are finally “due” and will shrug off the choker narrative.
The numbers suggest things could go either way.
Like the other Western Conference matchup, the Predators and Sharks were very close in terms of points during the regular season, second half, and last 10 games. Not surprisingly the Sharks enjoyed a +2.8% power play advantage over the offensively challenged Predators. Both teams had a decent but not great penalty kill, with Nashville (+0.7%) enjoying a slight edge, and they were nearly identical on ESVA Fenwick % (52.9% vs. 53.0%).
With the Ducks and Kings eliminated, the Sharks have suddenly leapt into the enviable position of being the Western Conference favorite—at least according to our model.
If you happen to have occasionally watched some hockey over the past few years it would be hard to miss the fact that the Sharks have a ton of offensive weapons. But unlike the Blues, who seem content to trade big chances with their opponents, the Sharks dominated on shot quality during the regular season, generating a league best high danger scoring chance differential of +259, significantly better than the high-octane Penguins offense, which was second best at +168.
The Preds were no slouches either (+132), but as San Jose showed in the first round, when it made 2012 playoff hero Jonathan Quick look like a call-up from Minor Peewee, this is an offense to be feared.
The differentials don’t tell the full story here though.
San Jose generated its advantage based on its league leading ability to generate high danger scoring chances, a total of 1,117. Nashville, on the other hand, was best in the league at suppressing high danger scoring chances, allowing only 730.
If you’re looking to test the old “defense wins in the playoffs” argument, this is your series.
Goaltending may also play a role here.
Nashville’s Pekka Rinne has been abused for some time by the analytics crowd, who routinely note that a guy who faces so few tough shots should have a much higher than average save percentage overall. Instead, Rinne continued a downward trajectory that’s been going on for a while, posting a .908 save pct. this season—a long way from the lofty .930 he managed in 2010-11 (when the Preds were only middle of the pack defensively).
The regular season suggested that Rinne could be either washed up or not as elite as some people thought he was.
TV commentators may have gushed over Rinne’s final two games of the Anaheim series, but overall his first round performance (.915) was fifth-best among the eight remaining starters, only slightly better than that of San Jose's Martin Jones (.912). Still, that middling performance was a significant jump from where Rinne was all season.
Based on regular season performance our model gives Jones a 0.9% advantage on xSV%. But it’s possible that number is an overestimate if Rinne is able to replicate or improve upon his first round performance.
One oddity in this series is that although our model gives San Jose the edge overall, out of the eight possible outcomes for the series it rates a Nashville win in seven games as the most likely (a 20.0% probability).
If the series is a short one, the likeliest reason will be that San Jose’s offense has pounded Rinne and the Predators into submission. But if Nashville’s elite blueline is able to contain San Jose’s offense and grind out a seven-game affair, it may pull off yet another upset.
Our revised Cup odds are below. If you want to follow Ian’s daughter’s picks for this round, you can see them here.
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
Dr. Mikal Skuterud (@mikalskuterud) is a professor of economics at the University of Waterloo.