A wave is quietly rolling over the world of hockey. Inspired by similar movements in other sports, many NHL teams have come to embrace analytics, the objective and systematic study of sports, often involving the use of applied mathematics. In the 2014 offseason, at least a dozen analysts were hired by various teams, which led comedian Sean McIndoe (aka Downgoesbrown) to refer to it as the “summer of analytics.”
Origins
Other sports have embraced objective analysis for quite some time. Baseball researchers discovered that many of the traditional measurements of player quality – batting average, runs batted in, etc. – actually weren’t that great at reflecting a player’s performance, and they began to develop their own statistics. Soon, researchers started using the new statistics not only to study the past, but also to evaluate players of the present. Billy Beane and Bill James would make a name for themselves in the early 2000s MLB, for the Athletics and Red Sox respectively, the latter being part of the 2004 squad that broke an 86-year championship drought. In 2011, Roland Beech won the NBA championship as a consultant with the Dallas Mavericks.
By comparison, hockey has historically not been great at stats-keeping. Even so simple a stat as goalie save percentage was not officially tracked until the 1980s. To be fair, unlike baseball, which proceeds as a series of discrete plays around a number of discrete locations (the four bases), the more fluid game of hockey is played on a continuum in space and time (i.e. changes on the fly every minute). A good analogy for the baseball/hockey differences would be the difference between a turn-based game like chess, and a real-time game like Starcraft – you’re going to have a lot easier time analyzing plays in chess.
The arrival of the Internet caused a revolution. Around the turn of the century, NHL scorekeepers began to record the exact times, not only for goals, but also for shots, attempted shots, faceoffs, and line changes, posting them online in real time. All of a sudden, it was possible for anyone with access to the Web to acquire NHL play-by-play data; many hockey fans began looking for trends in the data. Around 2007, some of these fans started sharing their discoveries on hockey blogs, beginning the hockey analytics movement in earnest. Over the next few years, quite a bit of progress would be made in improving our understanding of hockey.
Findings
One of the major discoveries made by these analysts was that teams get shots on goal a lot more consistently than they get goals. In fact, looking at shot differentials (shots taken minus shots conceded) was actually a good way of determining the strength of a team and predicting future results. This spawned a pair of shot differential stats now referred to as Corsi and Fenwick (after their developers) – the difference is that Corsi counts shots which get blocked as shot attempts, while Fenwick does not. Fenwick differentials have also been shown to be a better predictor of playoff success than regular season standings or home-ice advantage – for instance, the Kings and Blackhawks were the top two teams in terms of shot differential last season despite finishing 5th and 6th in the Western Conference, and the Kings ended up winning the Cup with the Blackhawks being the closest team to knocking them out.
Little correlation was demonstrated between fighting or physical play and the success of a team, which has led many analysts to proclaim that these traits may be overrated. While the Kings have a strong forecheck which leads to them hitting a lot, Chicago plays a style built around entering the offensive zone with possession (which, with all their skill players, they’re very good at). Both teams have found success.
Other aspects of the game which analysts have been exploring are the quality of shots faced by a goalie (to no one’s surprise, it’s more challenging to play goalie for the Leafs than most other teams), the advantages and disadvantages of dump and chase versus carrying the puck into the zone (carrying is better if you can avoid turnovers), and the best time to pull your goalie when you’re down a goal (it’s earlier than the final minute).
New Hires
Over the past few years, some NHL teams had begun to do similar work to the amateur analysts. However, the summer of 2014 would bring a hiring spree. In July, New Jersey Devils hired Sunny Mehta as a consultant. Mehta, a former professional poker player, had done a lot of research into shot blocking.
In Edmonton, Tyler Dellow was hired as a consultant. Dellow had, for several years, trashed the team’s management for failing to improve the team, and had done an in-depth study of the defensive game (or lack thereof) of the Oilers stars. The Maple Leafs named Kyle Dubas – former Soo Greyhounds GM, who had been known for incorporating analytics findings into his team philosophy – as their assistant GM. Gabriel Desjardins, one of the pioneers of the Corsi stat, and Eric Tulsky, who had done research into score effects (the tendency for a team in the lead to go into a defensive shell), also received employment from NHL teams.
Evaluation
The analytics movement has many detractors. Sometimes analysts are seen as trolls who like prophesying doom for rival teams, but there are legitimate criticisms. The new statistics being developed in hockey are still in their infancy, and there are a lot of kinks still to work out. For example, Corsi simply treats all shot attempts the same way, regardless of the location or quality of the shot. In a recent interview, Red Wings head coach (and two-time gold medallist) Mike Babcock reported that he was not too keen on the new stats. It should be noted that Babcock has for years been a proponent of a skill-based, puck-possession style, which is very similar to what the hockey analysts have been promoting. Perhaps the current research has only provided evidence for what coaches already know. I mean, it’s really not that hard to figure out that it’s better to have the puck than not … But with the pace of research being done, this might not be the case for long. Perhaps some teams have already made some ground-breaking discoveries, and are simply keeping quiet about it.
We might just be on the threshold of a revolution in how teams understand the game of hockey.
Leave a Reply