A Clustering Approach to Analyzing NHL Goaltenders’ Performance

Ice hockey is among the top 10 sports in the world by global popularity, and the National Hockey League (NHL) is one of the major professional sports leagues in United States and Canada. In the NHL there are 32 teams, 25 in the U.S. and 7 in Canada. In ice hockey, the goaltender, also known as the goalie, is one of the most important players in the game. The result of the game greatly depends on the performance of the goaltender. One of the most important statistics of the goaltender is save percentage SV% (calculated as the number of saves divided by the total number of shots attempted on the goal). In spite of the goaltender being a key player in the game, there are shortcomings in existing methods of ranking goaltenders, as these methods do not comprehensively capture the performance of the goaltender. This paper proposes the use of clustering methods from machine learning to compare performance of NHL goaltenders by using SV% and to look for patterns in their performance.