Using Data Analytics in Hockey Player Talent Identification

Abstract

Goal for the research is to identify relevant data and features which can help the team to find rising future talent to build a data and analytics model and simulate the analysis with real data. The aim and main objectives are to create both data and statistical models for analysing a hockey game data in relation to the defined indicators for identifying better than usual player performance and to run the analysis with real life data.

Data analytics in professional sports can still be seen as a relatively new concept. In professional hockey leagues, like Liiga or NHL, a lot of data related to games are collected, analysed, and even made publicly available.

This research is conducted by analysing the previous research documentation, articles, and results, by collecting information from trusted web sites, blogs, and articles and using them to define the data, features, and statistical models, and building data model and data pipeline and finally executing an example analysis with the build models and collected real data. This research has quantitative and qualitative research parts. Qualitative parts goal is to define the theoretical framework and the quantitative part includes building a data pipeline and visualizations, and running the analysis based on the indicators and model built in the qualitative part. There are numerous statistics and metrics which all try to explain player performance. Data is collected from games and made available via commercial or public channels, like Wisehockey for Finnish Elite Hockey League, Liiga. In analysing player performance, focus should be put in what happens when player is on ice. With analysing data from games and by using machine learning to detect anomalies in players performance can help teams in player recruitment decisions. Machine learning based predictions and analysing video data from games can help teams in future not only on recruitment but forming winning composition and identifying game patterns during game. The key to realizing these benefits is access to relevant and trustworthy data.

Future research should focus on using novel methods like machine learning to bring full benefits from data and analytics.