Optical Tracking of NHL Hockey Games: A Big Data Approach

• Activity Number: 144
• Type: Invited
• Date/Time: Monday, August 4, 2014 : 10:30 AM to 12:20 PM
• Sponsor: Section on Statistics in Sports
• Author(s): Marc Appleby*+
• Companies: Powerscout Hockey

Abstract

Hockey is a game with continuous flow and complex within-team and opponent-team interactions. Therefore, modeling the underlying process of winning is difficult relative to other sports. Previous attempts to evaluate players by Gramacy, Jensen and Taddy (2013) and Thomas et al. (2013) that analyze a player's marginal contribution to goals scored for and against have used event data that is captured by the National Hockey League (NHL). While some of the results obtained in the previous attempts align with the mass consensus for elite players, usable results across the group of non-elite players have been difficult to obtain for a number of reasons.

Recently, a new set of data captured using Powerscout Hockey's player and puck tracking technology has become available which, for the first time ever in hockey, almost every aspect of the game is being measured, including player speeds and accelerations, passing efficiency, puck possession, and zone time for example. This dataset is significantly more granular compared to the previously used event data captured by the NHL, and also contains information relating to assessing player performance not previously possible to measure.