Abstract

This essay seeks to describe the sport of bandy using sports analytics while working with the Swedish Elitserien team Sirius Bandy. The task entails data collection, finding suitable analytical models and presenting their results in a helpful manner. The essay will also seek to confirm or deny some hypothesis about the sport of bandy. An array of Python files and programs was created to help the user collect data from a televised live bandy game, create statistical models from the collected data and then automatically generate two types of presentations that presents the findings from these models. These presentations take the form of a Game Report, which can be created both in the half-time of a bandy game as well as quickly upon its conclusion, and as a Season Report which can be used to analyze more than one game. The code, along with two example reports, can be found in the following GitHub. For more information on how to use the code see the README.md file.

When attempting to understand the sport of bandy with the help of these models some conclusions were made regarding the importance of shooting percentage, shots on goal and shot attempts. It has also been shown that duels between two or more players play a somewhat important role for the games’ outcome but require more work to fully comprehend. No relation has been found between ball possession and the outcome of games. Some key aspects which differentiate play indoors and outdoors have also been identified, most notably an increase in the total number of duels and long passes. These differences are amplified when the ice quality outdoors worsens. More work is needed to fully comprehend the sport of bandy, and this is only a first step.