Artificial Intelligence in Hockey Analytics

Hockey is a sport that involves constant motion among 5 skaters per team on the ice. Each goaltender typically stays put. Analytics is the study further into the statistics of a game to further make predictions. Analytics has been a rapidly developing field since the late 1900s with many notable figures using it; one that I like to mention is Billy Beane of the Oakland A’s, featured in the book and later movie, “Moneyball.” Using sabermetrics, he was able to build a baseball team that was successful due to the use of undervalued players. He analyzed stats the game had, with data points already measured.

Hockey isn’t so easy to evaluate. There are many moving parts, without any sense of turn-based action like baseball or football, where each play kind of resets. Tracking the players and shots based on location has been very difficult. Add in the association of each player per team, and you have many uncollected data points. With the addition of cameras around the rink, analytics professionals can use AI systems to track players’ every movement on the ice with relation to their teammates and the puck. This makes calculations for expected goals per player and per location much easier to record and determine thanks to new data points that were previously uncollected and hard to collect.

The use of analytics since the 90s with Billy Beane was just the beginning, and the use of AI and other future systems will revolutionize each sport.

Article: https://medium.com/syncedreview/ai-powered-hockey-analytics-a-game-changer-8534b2e263aa

6 thoughts on “Artificial Intelligence in Hockey Analytics

  1. I don’t think using analytics in hockey will help too much. In baseball, its all about the batter and the pitcher and so analytics can be used for defensive positioning. Football gets more complicated as there are more moving parts and schemes to consider. However, hockey is a fluid game that does not have set plays or batting sequences. A big part of hockey is team chemistry especially in the playoffs when the head coaches are switching up the lines to try and find that spark. Analytics could possibly help in making those decisions, but in the end I think it will mostly come from playstyle which analytics won’t be able to show the patterns in players playstyles very well for a human to recognize. To me the biggest stats that will help is play time and points totals for GM’s to decide on who to pick up from free agency or trade for.

    Link: https://blogs.oracle.com/bigdata/how-hockey-is-embracing-big-data-and-analytics

  2. I am very interested in this article. I am an amateur Texas Hold’em player. At first, people thought that AI could not crack Texas Hold’em because poker is a game with incomplete information. But it turned out that the best Texas poker player in the world is AI. More and more table games have been slowly cracked by AI. Does this mean that these games will slowly disappear in the future? AI’s learning ability, computing ability, and ability to adapt to the environment cannot be underestimated. It is a level that humans will never reach. How to make good use of AI is a question we should think about.

  3. Artificial Intelligence has been taking over the world of sports in general. It can be used in many different aspects such as scouting and recruiting, training and performance analysis, maintaining health and safety, and broadcasting and streaming. In many modern day sports, one can use individuals performance data to help make projections about them. We can use AI and historical data to make predictions about their potential before investing in them. This is a crazy feature because before artificial intelligence, predictions were very hard to make without the knowledge of previous performances. Artificial Intelligence can also identify opposing team’s patterns and tactics, therefore helping coaches devise detailed plans to leading their team to victory. There is no doubt that artificial intelligence makes sports predictions more certain and reliable, but we can never rely on this 100 percent. That is why I, and many others, love sports. No matter how much you look at the stats or charts, if someone has an off day or gets hurt, that affects the whole game.

    Source: https://www.forbes.com/sites/cognitiveworld/2019/03/15/heres-how-ai-will-change-the-world-of-sports/#702e1ff6556b

  4. Analytics is becoming more and more prominent in the entire sports world. I think it is so cool how they can use analytics to help their teams before better. They can see at what points in the game players perform the best and from what spots on the rink/court/field. I think in hockey analytics can be especially helpful because of how fast they game is without any pauses. With substitutions on the fly it is important that they don’t lose valuable time and possession while making the subs. They are also putting little chips into the pucks and into players shoulder pads to get every piece of data possible about the game. Increasing use of AI in all sports is definitely going to progress the sports and make teams and players more and more efficient. It is exciting to think about.
    https://blogs.oracle.com/bigdata/how-hockey-is-embracing-big-data-and-analytics

  5. The NFL is also doing a similar thing with stats. Amazon Web Service is using machine learning in order to provide their “Next Gen Stats”. They use machine learning in order to calculate things such as catch probability, an estimate on the yardage a play will gain, and more. While I don’t think this adds much to watching the game, because you get to see what happens on the play. I can see how these stats could help the teams playing. They could use these stats in order to choose their plays. This advances the game even more, and adds another strategic way to playing the game.

    Sources:
    https://aws.amazon.com/nfl/

  6. This is a really interesting article, and I think that what is being discussed can definitely be seen as true in all of the other sports. It is true that in baseball, the Athletics and Billy Beane used sabermetrics to find a number of undervalued players in the league, and acquired them for cheap so that they could play against competition that was always spending a lot more many than them on players. I feel that this is really only a system that will 100% work for baseball, as it is the only sport that does not have a salary cap, meaning that each team can spend different amounts on the players on their team every year, and the teams in bigger cities usually are able to spend more money. I feel that there can definitely be comparisons in the methodology used by the Athletics in “Moneyball” to the point where teams should always be looking to sign undervalued players so that they can spend money in other areas to make the team better. In a sport like hockey, where they already have a salary cap and a standard set for contracts for players, I feel that it would be incredibly hard to find players that other teams would not also already be interested in so that you could sign them for a lot less money that what their performance on the ice actually says they should earn. It will be really exciting to see this in action in other sports, and if the methodology is really the same.

    Source: https://puckprose.com/2019/05/03/moneyball-type-analytics-will-never-work-professional-hockey/

Leave a Reply