In the 2008 playoffs, the Dallas Stars limped into the fifth seed in the Western Conference with a 3-5-2 finish and 97 points.
As the third-best team in a strong Pacific Division, the Stars had drawn the defending Stanley Cup champion Anaheim Ducks as their first-round opponents.
The Ducks finished five points ahead of the Stars, but they also won their last three games of the season and were 8-2-0 in their last 10. Moreover, Anaheim had home-ice advantage and was 28-9-4 in its barn, second only to Presidents’ Trophy-winning Detroit.
All signs seemed to point to an early exit for Dallas, yet somehow the Stars managed to surprise the Ducks and earn a 4-2 series win.
Up next were the San Jose Sharks, who had won the Pacific and owned the second-best record in the league. Once again the Stars were victorious.
It wasn’t until they faced off against the Red Wings that Dallas’ “lucky run” came to an end.
But were the Stars actually lucky?
By one measure, Dallas was the best team in the Pacific that year. Yes, the Stars finished third (and significantly behind San Jose) in their division, but their goal differential, was third best in the NHL (plus-35) and second in the West only to the Red Wings.
Goal differential alone doesn’t tell us how strong a team is. The reason for this is that some teams have tougher schedules than others.
Thanks to the folks at hockey-reference.com, we now have a number that addresses this problem. It even has a name: simple rating system (SRS).
SRS is – as far as I can tell – an underappreciated metric because the glossary on hockey-reference.com is a little thin on details as to how it’s calculated, and the math is a little complicated.
But in simple terms it works like this:
Each team is assigned a ranking based on its goal differential per game as well as the strength of its schedule, which takes into account that teams playing weak opponents will have better goal differentials than ones playing tough ones.
For example, Tampa Bay, which is third in the league in goal differential this season (plus-48 through Saturday’s games), is assigned an initial ranking of 0.67 (48 divided by 72). But because they play in a weak Atlantic Division that includes juggernauts in the Connor McDavid/Jack Eichel sweepstakes such as the Leafs and Sabres, their SRS is adjusted downward to 0.58.
Now remember those 2008 Stars? As it turns out, they were second in SRS that season (0.54), behind Detroit (0.93) but ahead of both San Jose (0.47) and Anaheim (0.30).
Before you conclude this is an aberration and Dallas had some ethereal quality that pundits usually lean on after the fact (heart, grit, experience, mojo), we looked back at all 105 playoff matchups since 2008.
We found that if you had used the difference in each team’s SRS to predict each series, you would have gotten 70 (66.7 percent) of those matchups correct; if you looked at point differential, you would have gotten 61 matchups (58.1 percent) correct.
This suggests that when deciding whether a team is a favorite, pundits should be looking at which one has a higher SRS rather than which has more points.
Making predictions isn’t an easy business, and when doing it, you have to be honest about your odds just by guessing at random.
When I ask my preschooler (whose main interest in hockey stems from knowing she gets to extend her bedtime) whether the Penguins, Blackhawks, Red Wings or Rangers will win the Stanley Cup this year, as long as she doesn’t answer “Blackhawk Red Wings” (which occasionally happens) I’m giving her a 25 percent chance of looking like Nostradamus out of the gate.
Mainstream pundits have an easier job; just by flipping a coin they’ll be right half the time, and the other half nobody will remember what their predictions were, anyway.
As we approach this year’s playoffs we’re already working on doing better than the 66.7 percent SRS gives us.
More on that in the weeks to come.
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 litigator at the law firm of Armstrong Teasdale, former high-stakes professional poker player, and Harvard Law School graduate. Please visit us online at http://www.depthockeyanalytics.com/.