To overcome the shortcomings of simple metrics for evaluating player performance, recent works have introduced more advanced metrics that take into account the context of the players` actions and perform look-ahead. However, as ice hockey is a team sport, knowing about individual ratings is not enough and coaches want to identify players that play particularly well together. In this paper we therefore extend earlier work for evaluating the performance of players to the related problem of evaluating the performance of player pairs. We experiment with data from seven NHL seasons, discuss the top pairs, and present analyses and insights based on both the absolute and relative ice time together.
© Copyright 2019 Machine Learning and Data Mining for Sports Analytics. MLSA 2018. Lecture Notes in Computer Science, vol 11330. Veröffentlicht von Springer. Alle Rechte vorbehalten.