Hockey analytics is the systematic collection, analysis and interpretation of data to gain insight into player performance and team tactics. Advanced tracking technologies record a vast array of data points during games including shot locations, skating speed, passing accuracy and much more. The application of these insights helps coaches and trainers develop a deeper understanding of their players’ strengths and weaknesses, as well as the trends in the game itself.
This information can then be used to improve training programs, identify areas for improvement and accelerate the development of individuals and their overall game. Ultimately, the use of these tools can lead to greater success and enjoyment for both teams and their fans.
While the use of analytics in hockey has grown, it is still a relatively new field. As such, there are still some misunderstandings that can hinder the implementation of this information to the fullest extent possible. Despite this, the benefits of hockey analytics are clear and the importance in using these tools is increasingly being recognized.
There is a growing push in the NHL to have analytics as a core part of coaching and player evaluation. Some of this stems from the belief that numbers are more accurate than the old eye test, while other managers and coaches want to be able to better evaluate a player and their role on a team.
As a result, teams are investing heavily in advanced puck and player tracking. This allows coaches and trainers to cut live video of players during games, tagging breakouts, scoring chances, faceoff wins and more. This allows them to evaluate players in a more comprehensive and precise manner than ever before.
Some of the most common advanced statistics in hockey include Corsi and expected goals (xG). These metrics are calculated by analyzing past shot attempts and then assigning a probability that each individual chance will become a goal based on the shot location, type and quality as well as the opponent and other factors. There are a number of different models which can be run to come up with an expected goals total for each team and player. Some of these can be quite complex and the math involved is a bit beyond most people’s comprehension, but there are some good resources available to learn more about them if you have the time to dig into it.
The problem with these stats is that they are often used in isolation and don’t take into account the quality of competition or teammates a player is facing. For example, a 3rd line forward might have great Corsi numbers because he is playing 15 minutes a night against the opponents 4th line and not their 1st lines. In this case the stat is being overly skewed by puck luck. This is why larger sample sizes are more reliable, as they tend to normalize for this phenomenon.