Earlier today I attended the Pittsburgh Hockey Analytics Conference at Carnegie Mellon University, hosted by Andrew Thomas and Sam Ventura – the founders of the analytics website WAR-on-ice.com.
The day touched upon a variety of interesting topics to the NHL and hockey community at large including introductory remarks explaining the history and current state of analytics in hockey, the adoption of analytics by mainstream media, and where we should expect analytics to trend in the future.
Here are five interesting things we can take away from today’s discussion:
1. There is a growing appetite for data driven analysis amongst hockey fans – as attested by the 200-plus people registered for the event, many of whom came in as analytic neophytes. Cross pollination amongst sports will also be helpful for improving methodology, approach, and improving the dialogue between management, coaches, players, and analysts.
2. Shot Suppression really is key to teams being successful in the regular season and in the playoffs – AKA defence wins championships. The widely assumed relationships between shot attempt generation and suppression are almost nonexistent – we need to account for both as distinct abilities, and they don’t seem to be of equivalent value when it comes to winning. This largely explains the dominance of teams like LA, Chicago, and St Louis in recent seasons.
3. Similarly – seeing data collection improve in hockey at lower levels (junior/college/minor leagues) may enhance implementation at those levels. A presentation by Kody Van Rentergem on the implementation of statistical video analysis with the Robert Morris University Men’s Ice Hockey program showed that similar results are being observed in the NCAA and the NHL. Implementation and improvement using analytics is an organic and ongoing process.
4. We may be able to tease out more interesting, underlying information on zone transitions eventually using Zone Transition Time (ZTT) in the not distant future thanks to the efforts of CMU PhD Sam Ventura. His work is in the early stages, but shows promise. Another new effort to keep an eye on is the passing efficiency and network analysis work being done by Ryan Stimson and others.
5. The future of hockey analytics is in statistical modelling of team and player results based on the information we have. Literally trying to predict how players are going to perform in the future. Getting more data from sources like video tracking or RFID tags would help with these efforts, but assessments can be improved with the data that is already available. Accounting for context as much as possible can also improve or analysis.
The day was very informative for all involved. As more events like this pop up around the hockey world, I’m sure the analytics discussion will improve, grow and progress forward.