Analysis of the relationship between physical parameters of hockey players and their performance statistics in the world's top leagues

This bachelor thesis focuses on the analysis of the relationship between hockey player’s physical attributes (age, height, and weight) and their performance statistics in five of the world’s most prominent hockey leagues: the North American NHL, the Czech Extraliga, the Swedish SHL, the Finnish Liiga, and the Swiss National League. The aim of the thesis is to determine whether there is a correlation between a player's physical characteristics and their performance across different phases of the season (regular season and playoffs), and to compare differences across the respective leagues. Machine learning methods were used for the analysis, specifically linear regression, neural networks, random forests, and gradient boosting algorithms. The dataset was created using web scraping, then preprocessed and extended with derived variables. A custom normalized player performance rating metric was also constructed, enabling objective comparison across leagues. The results indicate that age was the most significant predictor of player performance. In contrast, height had minimal influence, while weight showed slightly higher importance— especially among defensemen. In the NHL, player salary also emerged as a strong predictor. The models achieved higher predictive performance in the regular season than in the playoffs, likely due to the smaller data volume and higher variability of player performance during the shorter play-off period. In conclusion, this thesis provides new insights into the factors influencing player performance and highlights different trends between North American and European hockey. The findings contribute to a better understanding of the relationship between physical parameters and player performance, particularly from a data analytics perspective. In practice, they may serve as a supplementary tool for scouting or for comparing players across leagues.