Idea in Brief
Professional hockey teams are increasingly investing resources into analytics personnel and teams to gain strategic insights for players and coaches. The hockey team’s data-based insights are only as good as the quality of the data available. Hockey game stats are recorded and collected in less than standardized ways – humans in various arenas recording shots, hits, etc. subjectively – and are prone to significant errors.
Statkeeper will apply Computer Vision and Deep Learning techniques such as continuous pose estimation on video feeds of hockey games to automate the tabulation of game stats to significantly standardize and improve the quality of the data.
Many small-time junior leagues don’t possess resources to record stats. This impedes the ability of analytics teams of NHL clubs to determine the draft potential of young players from unknown markets abroad. Statkeeper in its inception will focus on generating hockey stats for such leagues to provide significant value add for NHL clubs.