On Saturday in San Jose, the tech capital of the world united with the world of hockey for the first time publicly in the form of player and puck tracking.
All kinds of information flooded the screen for those tuned into the NBC website stream, including labels for the players, ice time clocks, speed, distance skated, a faint trail on the puck reminiscent of the infamous Fox “GlowPuck,” and more.
It might have been a bit much for the average fan, but that wasn’t the point. The point was that after years of limited testing beginning at the 2015 All-Star Game in Columbus, the NHL was ready to show the world the technology that will be introduced to all teams during the regular season starting next year.
The effects of having this data will have a groundbreaking impact on how the game is watched and analyzed for years to come.
Well not right away, but soon.
While this technology lays a strong foundation for the next era of the NHL, evolution in actual player analysis will still take time.
Current public statistics and advanced evaluation methods like Corsi, scoring chances, and expected goals are just more detailed ways of interpreting data the NHL has given us for years. They’re more descriptive than things like shots-on-goal and +/-, but they still lack some detail.
For example, for a long time, Corsi was referred to as “possession” by hockey analysts. Some still refer to it that way. The idea being that if you had a shot attempt, you had to have possession of the puck. As it turns out, that’s not entirely accurate. A team that has possession of the puck that is cycled around the boards but doesn’t get a shot attempt off still has possession. And the longer you hang onto the puck, the better—which is why Corsi falls short of being an accurate measure of which team has greater puck possession.
This player and puck tracking technology solves that dilemma by giving us a true measure of possession. And with it, entirely new models of player and team performance are waiting to be built; new scoring chance metrics, new expected goal metrics, and more.
So why can’t we have these things now? For starters, much of the data that will be available at first won’t be available for public use. ESPN’s Greg Wyshynski’s fantastic look into the development of the tracking system details an agreement between the NHL and the players union on how the data will be used at first:
So those doing wonderful public work in the hockey analytics space like Micah Blake McCurdy, Sean Tierney, Dom Luszczyszyn, Emmanuel Perry, and more won’t be able to significantly update their models at first.
More importantly is that many NHL front offices themselves won’t change their evaluation methods significantly in the first few years.
“But if the info is propriety for the teams, why won’t they?”
I’m glad you asked! Raw data means almost nothing without the right people analyzing it. Having hundreds or even thousands of lines of data showing how fast a player skated is useless on its own. Having a trained data scientist who can comb through large amounts of numbers and build them into models with in-game context to provide meaningful analysis is the other half the new equation that is player evaluation.
Some teams, like the Toronto Maple Leafs and the Carolina Hurricanes, have been devoting resources to building out analytics departments staffed by these very qualified people over the last couple of years. These teams will have a significant head start in new and more accurate player evaluation methods.
Others, like the Edmonton Oilers, will be at a competitive disadvantage until they commit to investing in this new discipline.
With the introduction of this standardized data provided by the league, hockey is now at a similar inflection point as baseball was in the early 2010s with the development of the PITCHf/x system and its eventual successor, Statcast. PITCHf/x, introduced in 2006, used a camera-based system to measure aspects of a thrown ball like speed, spin rate, amount of break, etc. Statcast, introduced in all ballparks in 2015, applies that idea to pitches as well as player movement and in-play ball movement using both high-speed cameras and Doppler radar.
The build-up had been a long one, and we saw the potential of that data on full display when the Chicago Cubs, with one of the largest analytic departments in baseball, broke their 108-year championship drought in 2015 against the Cleveland Indians, another heavily analytical team. The Los Angeles Dodgers have made back-to-back appearances in the World Series due largely to these analytics departments, and the Houston Astros have built what many argue is one of the best young teams ever to take the field with these data scientists.
Hockey, however, has a long way to go to get to this point, especially with the sport’s infamous resistance to change in most aspects of the game. It will likely take a Stanley Cup win from a team taking the financial risk of building out an analytics department to bring us to the tipping point.
That’s not to say that the implementation of this technology doesn’t mean anything. This is just the first yet extremely important step towards building the next generation of player evaluation methods that will give teams and, eventually, the public, truly accurate statistics to take the game to the next level.
Why does this even matter to you, the hockey fan? Analytics are credited with giving baseball its most talented crop of young talent in its 100+ year history. Players have new information on how to improve their existing skills, and teams are fielding more and more competitive teams playing at a level never before seen in the sport’s history.
Eventually hockey will reach the same level, and these statistics made possible by the player and puck tracking technology shown off at this year’s All-Star Game will usher the sport into a new and more exciting era.
It might take several years before this data becomes truly meaningful, but the ingredients are now there.
The only question now is which teams will hire the chefs to turn this into something delicious?