Beautiful minds have been influencing baseball for three decades, and there is mounting evidence that mathematics wizards are infiltrating the National Hockey League, exposing truths beyond the box score.
It isn't quite an invasion of the nerds - as witnessed in baseball, where the disciples of sabermetrician Bill James have populated every front office - but the field of hockey analytics, where advanced mathematics is used to explain occurrences on the ice, is expanding. The explosion of statistical analysis in basketball has forced NHL teams to take notice of the guy on his couch with a calculator and a formula, someone who may know when a coach has erred in his choice of faceoff men.
"With the salary cap, it's harder to find ways to be better than the team next to you," Minnesota Wild director of hockey operations Chris Snow said.
"We all have to find value elsewhere, outside of paying players."
Clandestinely, several NHL teams appear to be digging deeper, turning to mathematicians, statisticians and systems analysts to unearth clues. Not many wish to speak about their involvement, for fear of tipping off competitors, and no team would disclose specific areas of study.
But the Wild, Buffalo Sabres, Toronto Maple Leafs, Vancouver Canucks and San Jose Sharks are all believed to be working in analytics, according to NHL and online sources.
The Leafs and Canucks attended the Massachusetts Institute of Technology's fourth annual sports analytics conference this month.
Leafs general manager Brian Burke was a panelist, and his daughter, Katie, is a former conference chair.
In 2006, the Wild hired Snow, a former sportswriter who covered the Boston Red Sox and GM Theo Epstein's devotion to statistical analysis. The Sabres' involvement is so strong that one of their assistant coaches has a new statistical measurement named in his honour, and many believe the Sharks have been working with analytics for years.
"They've got a lot of computers in San Jose," one source said.
Likewise, the community of online number crunchers is hesitant to speak of their involvement with NHL teams.
Mike Phillips runs Sydex Sports, a software company based in Grand Rapids, Mich., that works with 10 teams and the league office. Phillips is busy synchronizing statistics with video, but when contacted for an interview, Phillips said: "we work closely with teams, and as a policy, shun outside publicity."
Several sources from the underground of hockey math told stories of contracted work for NHL teams, to casual e-mail inquiries from team personnel. One Southeast Division team was recently fishing for some basic information, suggesting it has just caught on to analytics. Thirty minutes into one interview, one mathematician said: "I don't mean to deceive you. I do actually work for an NHL team."
Gabriel Desjardins is an engineer based in San Francisco. But the Winnipeg native, who has degrees from Queen's University and the University of California-Berkeley, has worked with teams in all four professional sports leagues. Hockey is still waiting for its defining thinker, such as James, and its defining statistic, such as on-base percentage, but Desjardins says his work is positively reinforcing management decisions in hockey, as opposed to the statistical revolution that changed baseball.
"In baseball, you had Neanderthals working as GMs, and they were ignorant of a lot of things that helped their team win," he said. "Hockey teams, by and large, do know what they are doing. Coaches have a very good understanding of their players."
Snow said that NHL teams fall into three camps on analytics: devotees, recent experimenters, and non-believers.
Hockey is clearly less quantifiable than baseball, but basketball analytics, led by Houston Rockets GM Daryl Morey, an MIT graduate, has created a model from which to work. Both sports are five-on-five games of free-flowing action.
Some analysts say the sports cannot be purely compared - "in football, baseball and basketball, you take turns playing offence and defence. In hockey, you're playing offence and defence 100 per cent of the time," said Toronto-based mathematician Alan Ryder. However, NHL teams are nonetheless looking at new models for measuring players, frameworks that isolate individual actors and individual plays, and judge outcomes and responsibility more accurately than official statistics such as plus-minus and save percentage.
Number-crunchers assign values to questions such as: who is on the ice when goals are scored? At what point in the game? Where are they on the ice surface? What are the conditions (even strength, power play, short-handed)? Where was the faceoff? Who won it? Against whom?
"Do we know how to create the ideal line yet? No," Snow said.
"[But]if you don't give yourself every opportunity for as much information as possible, there are going to be regrets."Report Typo/Error