Year: 2016

Yesterday, Babson College in Boston hosted their first ever “Analytics On Ice: The Long Change” hockey analytics conference. “The Long Change” is of course a double entendre for hockey fans, with the implied second meaning being that the game is just beginning to embrace analytics, later than most sports and with a long way to go. Hockey is a long way off from being where it could be from an analytics perspective, and many still believe that hockey is a game for the eye-test rather than numbers and formulas, but there is a definite and undeniable role for analytics in the future of hockey. With the advanced statistics movement in hockey finally starting to gain steam, collaborations of the best minds in the topic have become common.  Bringing together experts in statistics, those involved in the hockey industry, students, and fans alike to share ideas and research and learn more about the game and the numbers behind it will help to increase awareness and understanding of the movement in an effort to work toward a greater understanding of the inner workings of hockey and the broad applications of analytics to the sport.

Whether Corsi and Fenwick sound like a foreign language to you or not, anyone could have learned some fascinating information about hockey analytics on Saturday. Mixed in with statistical software tutorials and paper presentations on topics like the value of stay-at-home home defenseman, the performance of teams dressing seven defensemen and eleven forwards, forming international teams, and more, were two excellent presentations on game-changing analytics topics.

The first talk, by St. Lawrence University professor Michael Schuckers, explored new ways of evaluating goalie performance using analytics. A veteran in the field, Schuckers did extensive research into the shortcomings of traditional goalie statistics and focused on how to best determine the value of goal tending. Schuckers explained that Wins and, to a lesser extent, Goals Against Average are not fair determinations of a goalie’s success. Wins are of course totally subjective in that they are determined by total team performance, and not just how well the goalie does. The example used was that of Pekka Rinne, who finished sixth in wins in 2015-16, while posting a save percentage that was well below average. Goals Against Average is also a function of team success, as a keeper allows goals based on how many shots the team in front of him is giving up. Thus, Schuckers focus was instead on Save Percentage, and how to further draw value from a goalie’s likelihood of making a save. The future of the stat in analytics is breaking down save percentage into types of shot. Using variables like location on the ice, angle, distance, and of course type (wrist, snap, slap, backhand, tip, wrap-around), a goalies true value can be determined. Save Percentage can be distorted when better goalies are facing more difficult shots, resulting in a lower percentage than inferior competitors. Schuckers suggests that with more accurate recording of shots, the stat can be more accurately expressed in an adjusted form based on either comparisons against the average keeper or against an average distribution of shots. Other interesting points in the presentation included the idea of weighing rebound rate (especially as a function of shot types) into goalie valuation, and the concept of the “royal road”, the imaginary line that runs down the middle of the net, which analytics show greatly effects scoring chances if a pass or player crosses the line just prior to a shot.

The second discussion was with renown hockey analytics expert Rob Vollman, who talked about the most important part of analytics, which is how to actually use it effectively in team building. In his recently-released book Stat Shot, Vollman put together a system for evaluating players, not just on their own ability, but on their relative value to other available players and as a function of putting together a roster with many different limits and rules. With variables such as the NHL salary cap, minimum and maximum contract values, entry-level contracts, and free agency rules, team building is not as simple as just taking all of the best players. Vollman has developed a system of evaluating players based on their value relative to their contract. When acquiring a player, their production has to be considered not as absolute but as relative to their cost. While Vollman went far more in-depth about formulas for ideal player cost-values as well as trying to evaluate a player based on a single metric, the crux of his presentation was that analytics can only be used effectively by NHL teams if statistics are just part of the equation, and market scarcity, acquisition costs, team structure and performance, and more are given their fair share of attention.

While the word “analytics” sounds scary, none of the above should come as too difficult to understand for the average hockey fan. It’s true that hockey is a very subjective game and there are some factors – like line chemistry for example – that can’t be quantified (yet). Scouting will always be crucial and “toughness” and “heart” will never be discounted, but a stronger understanding and application of analytics in just another tool for evaluating players and putting together rosters. Fans and teams alike should embrace the analytics movement and all of the promise that it brings. In the end, everyone wants their favorite team and players to do well, and numbers only help the cause. Consider attending a hockey analytics conference in your area if you hear of one, or take the leap and read up on some advanced metrics in your spare time. Hockey analytics is on it’s way to the forefront; don’t get left behind.