A Data Edge in the Faceoff Circle
- While professional hockey has been slow to adopt analytics, the National Hockey League’s planned adoption of puck and player tracking technology for the 2019-20 season will represent a “huge push forward.”
- A study employing advanced analytics offers a new way of measuring faceoff success, an aspect of hockey that is typically overlooked.
Hockey, like basketball and football, was once a game of Xs and Os. But a new generation of sports analysts, including Smith School of Business grad Nick Czuzoj-Shulman, is transforming it into a game of equations. For instance: WDBE = ∑[P(WinDirectionjisc)* E(Eventjisc)].
The formula relates to faceoff wins and subsequent game play based on variables such as whether a player was on his strong or weak side, the direction the puck went, and whether the win was clean (i.e., the centre sends the puck directly to a teammate vs. not clean, where a puck battle ensues or the teammate must skate to recover the loose puck). Czuzoj-Shulman outlined this formula to National Hockey League analytics experts and at least one NHL general manager (he won’t say who) at the 13th annual MIT Sloan Sports Analytics Conference in Boston earlier this year.
His presentation, “Winning Isn’t Everything: A contextual analysis of hockey face-offs”, and accompanying study highlighted results of more than 71,000 faceoffs during the 2017-18 NHL season. The study was conducted by Czuzoj-Shulman and his colleagues at the Montreal-based sports-analytics firm Sportlogiq.
Most NHL teams use Sportlogiq’s data analyzed by Czuzoj-Shulman, who earned his Master of Management Analytics degree from Smith in 2017. “I’m still pinching myself, to be completely honest,” he says of his current job. Czuzoj-Shulman joined Sportlogiq in May 2018 from Montreal’s Jewish General Hospital, where he was working as a statistical analyst. “Hockey and stats are two of my passions, and being able to combine the two has been a dream come true.”
A Focus on Faceoffs
While certain advanced analytics, such as Corsi – which measures shot attempts, including those blocked by a defender or that missed or hit the net, to determine offensive zone proficiency – have been embraced by NHL stats nerds, Czuzoj-Shulman believes faceoffs have been largely overlooked.
A typical NHL game features 60 faceoffs. But while their overall impact on goals and wins is minimal (it takes an average of 75 faceoffs to achieve a plus-one goal differential), Czuzoj-Shulman says “if teams can pull off the faceoff strategy they want to, and can set up a player for an open shot, there’s a ton of value there.”
A player’s faceoff win percentage is the traditional measure for whether an NHL coach chooses to deploy him in a key faceoff. But Czuzoj-Shulman argues that multiple variables can help determine suitability to take the draw. He cites an overtime game between the L.A. Kings and Boston Bruins early in the 2017-18 season. While the Kings’ star forward Anze Kopitar had won just eight of the 23 faceoffs (34.8 per cent) he took that night, coach John Stevens sent him out to take a key offensive zone draw against Bruins forward David Pastrnak (who had won his one and only draw) with just 0.9 seconds left in overtime.
Kopitar won the draw cleanly back and to the inside of the faceoff circle where teammate Tyler Toffoli immediately one-timed a shot past Bruins goalie Tuukka Rask for the winning goal. “[The Kings] had other options they could have gone with if they wanted to look at just the win percentage, but they looked past that and sent him [Kopitar] out there and won the game,” says Czuzoj-Shulman. “What more could you ask for?”
Clean Wins Are Gold
A defensive zone faceoff basically comes down to winning the draw at all costs (or, at the very least, not losing it cleanly). But Czuzoj-Shulman’s study presents variables that can impact offensive zone faceoffs – from the handedness of the player taking the draw, to what side of the ice the faceoff is on, to whether the win is clean, to where the puck is directed.
Sportlogiq’s analysis found that a clean win in the offensive zone led to a shot on goal or scoring chance 38.6 percent of the time, versus 30.3 percent for non-clean wins. But not all clean wins are created equal. Clean wins directed back and to the inside of the faceoff circle led to shot events 43.6 percent of the time, compared with 32.1 percent for non-clean wins. Clean wins were also directed to a teammate’s stick 25 percent faster than non-clean wins.
“When a player wins the draw cleanly, his team can execute drawn-up set plays with greater ease and will therefore have better chances of catching the defending team off-guard or out of position,” the study says.
Czuzoj-Shulman says his study did not produce an official ranking of faceoff prowess but notes a few players who stood out. For instance, Henrik Zetterberg and David Perron were two players during the 2017-18 season with faceoff win percentages of less than 50 per cent; yet the draws they won were often clean and thus of higher value. Lars Eller and Ryan Getzlaf were also not the league’s top faceoff men; yet they were adept at directing the puck to more advantageous areas of the ice for their teammates.
Hockey Embraces Analytics
Czuzoj-Shulman is part of a growing analytics movement transforming pro sports. According to a report in Forbes earlier this year, sports analytics is expected to become a US$4-billion industry by 2022. It’s not just hockey either. “I’m a huge Philadelphia Eagles fan, and the year they won the Super Bowl [in the 2017-18 season] they were heavily into analytics,” says Czuzoj-Shulman. “I think teams are getting smarter, and any competitive advantage they can go for, they want.”
Analytics is also having a profound impact on how sports are played. Experts say analytics is the prime reason for a two-decade rise in three-point shots in the NBA. In 1997, teams averaged 15 three-point attempts per game; in 2017 they tried 27. Hockey is a relative newcomer to the movement. The NHL’s so-called “summer of analytics” – which saw teams across the league draft analytics experts – didn’t occur until 2014, more than a decade after Michael Lewis’ book Moneyball helped sports analytics achieve mainstream recognition.
Czuzoj-Shulman says that the NHL’s planned adoption of puck and player tracking technology for the 2019-20 season will represent a “huge push forward” for analytics. According to the NHL, the technology will include 14 to 16 antennas in arena rafters; four cameras to support tracking; a sensor placed on the shoulder pads of every player; and 40 pucks containing a sensor inside for each game.
Czuzoj-Shulman is hesitant to say that his study – and advanced analytics in general – will radically impact how hockey is played. But “there will definitely be nuanced changes to the game that will hopefully add up to something bigger.”
This article appeared in the Summer 2019 issue of Smith Magazine.