This past Tuesday at 6:30p.m., I almost drowned in hockey, thanks to a talk given by Michael Schuckers, an associate professor at St. Lawrence University and a big shot in the world of sports analytics. Statistical analysis in sports has been thrust into the limelight in recent years, undoubtedly partly because of the movie Moneyball, which recounts the story of the 2002 Oakland A’s baseball team. The general manager (GM) of the team Billy Beane, had an unlikely rise to success due to a type of statistical analysis called Sabermetrics, a form of statistical analysis in baseball that measures player performance.
This talk was hosted by the Desautels Sports Management Club, a student group in McGill’s Faculty of Managment as a chance for aspiring scouts, agents, GMs, and Sabermetricians to get a little wisdom straight from a guru instead of digging through dry papers or arcane blog posts to learn the craft. The event held at the Bronfman Building, was well-attended, with almost every seat in the small conference room taken. As Schuckers began speaking, his affable tone soothed my apprehension, and promised to deliver facts with a little side of entertainment. Then I registered the significance of his words and they chilled me to the marrow: this talk wasn’t just about sports statistical analysis, it was exclusively about hockey! And everyone else in this room knew a hell lot more about that than I do! And we were about to run a mock draft!
I guess I knew that as a sportswriter in Canada, my general indifference toward hockey would be punished someday, but I didn’t think I would be thrust into the deep end quite so quickly. Fortunately, the audience was split into teams of three to choose players, and I was gifted with knowledgeable and understanding partners who made all the choices and delegated me to writing them down in my journalist’s notebook. The purpose of the draft was to try to pick the five most effective players possible, and our choices were rated based on Schuckers’ own analytical statistic, the Total Hockey Rating (THoR). THoR is, at least to the uninitiated eye, an incredibly complicated metric, which, in Schucker’s words, measures “every event recorded by the NHL, [then] assigns value to those events based on the probability that they will lead to a goal.” The number then takes into accountsuch factors as the home-rink effect, power plays, and the statistical reporting eccentricities at each stadium, to come up with a number that represents wins above replacement (WAR) for every player in the NHL. (Fun fact: blocked shots in New Jersey are underreported by 27 per cent compared to the league average!) WAR is a unit of measurement borrowed from Sabermetrics that is used to measure a player’s contribution to their team. I was surprised to discover that Sidney Crosby, the only current hockey player I’ve ever heard of, did not even lead his team in WAR, suggesting that he is not the most vital player on the Pittsburgh Penguins (although he does place toward the top of the league). Instead, according to Schuckers’ analysis, the most useful forward on the Penguins is Tyler Kennedy. Judging by the gasps of my fellow audience members, other players’ THoR ratings were surprising as well.
My drafting team did not win. In fact, we only picked a single player in the top tier of THoR ratings, with our lineup including Jason Williams, Zach Parise, Jake Muzzin, Jimmy Howard, and our one star, Patrice Bergeron. The top non-goalie players according to THoR (apparently, statistical analysis suggests that goalies are the most replaceable players on any team) are: defencemen Kimmo Timonen, and Drew Doughty; left wingers Ray Whitney and Alexander Steen, and right winger Patric Hornqvist.
Next we talked about the future applications of statistical analysis like this for hockey strategy. Right now, Schuckers is most comfortable using his numbers for player evaluation and personnel decisions, as the data does not yet exist for sophisticated gameplay analysis. He was clearly excited, however, at the prospect of the NHL adopting technology similar to what already exists in the NBA, where the positioning of the ball and of every player on the court is recorded 25 times per second. This unprecedented level of data has already led to changes in the way NBA basketball is played, most dramatically in shot selection, with teams focusing on either getting to the rim or hoisting threes and leaving the poor two-point jump shot behind. The stats seem to show that mid-range jumpers just do not have the same expected gain as long-balls or layups in most situations. The same data can influence the NHL: for example, the Los Angeles Kings (arguably the most dominant playoff team in the NHL) base their game around puck possession. This decision it’s influenced by the rise of the Corsi Number, another complex form of statistical analysis used in the NHL.
Schuckers used his ThoR data to say that : when NHL teams are losing toward the end of games, they should pull their goalies to start a six-on-five attack much earlier than they currently do. As Shuckers said, “Losing by four is the same as losing by two. The risk of getting scored on is outweighed by the advantage of having an extra man on offence.”
This talk of changing hockey strategy made me think about how the rise of statistical analysis has altered the way other sports I love more dearly have been played in my lifetime. As one audience member pointed out, baseball, as a reasonably static, one-on-one sport, which was the first to be affected by the statistical craze has gone through stages where previous pronouncements by statisticians have been proven wrong by the arrival of more data. Most prominent among these was the Sabermetricians’ mantra that ‘defence doesn’t matter,’ which probably lengthened the careers of power-hitters like Adam Dunn at the expense of more balanced players. Now that it is easier to measure defensive performance, suddenly defence matters again in baseball! Statisticians and the coaches and GMs who listen to them need to remember to not get too confident about their numbers, especially in the early stages.
I also worry that the increasingly stats-driven approach to strategy will result in a dull sameness between all the teams in a given league, reducing the lovely idiosyncrasies that made your team yours. It’s exciting to watch NBA players chuck threes all the time, but I also get a little nostalgic for the 76ers of my Philly youth, coached by the fundamentalist zealot Larry Brown. If you shot a three on that team, you had better have been either Allen Iverson, the untouchable superstar, or completely open with a teammate under the basket to rebound. Otherwise you got the death glare.
But I’m not a cranky traditionalist. Sometimes conventional wisdom is even more soul-crushing than math. I look forward to the magical day when American football teams start listening to the stats nerds and start going for it on fourth and short regularly, and I love that the Sabermetricians admire underhanded baseball pitchers, dismissed by the old guard as mere novelties. In any case, it is useless to argue with the takeover of statistics in sports. It is the way of the future, whether I like it or not. And judging by the mesmerised attention paid to Schuckers at this event, the Desautels Sports Management Club may produce some hockey statisticians not too long from now.