E-Thesis 111 views
A machine learning and data-driven assessment model for characterising and predicting a teams success in championship football . / LEWIS EVANS
Swansea University Author: LEWIS EVANS
Abstract
Measuring different styles of play (SoPs) that football teams can adopt and how they relate to their performance is crucial in analysing optimal play styles for the league in which the teams play. The first objective was to analyse all the matches in the English Second Division, the Championship, 20...
| Published: |
Swansea, UK, Wales
2023
|
|---|---|
| Institution: | Swansea University |
| Degree level: | Master of Research |
| Degree name: | MRes |
| Supervisor: | Brown, R |
| URI: | https://cronfa.swan.ac.uk/Record/cronfa65461 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Abstract: |
Measuring different styles of play (SoPs) that football teams can adopt and how they relate to their performance is crucial in analysing optimal play styles for the league in which the teams play. The first objective was to analyse all the matches in the English Second Division, the Championship, 2021/2022 season, to identify SoPs based on seven separate groups of metrics. Front Foot Defending, Defending the Goal, Build Up, Attacking the Goal, Physical Totals, Physicals In Possession and Physicals Out of Possession. Using Principal Component Analysis and K-Means Clustering, each team was clustered for each category of metrics. The Dixon-Coles Time Decay model was replicated for the second objective using match results from the 2018/2019 to 2021/2022 Championship seasons. This was then improved using StatsBomb’s Expected Goals model for the 2021/2022 Championship season, which created a rating for all teams within the league. Combining these allowed for the identification of the Championship being a demanding league with high possession and high pressing teams with good physical outputs performing the best. Most other teams showed a variety of SoPs with varying success. The applications of this study are (1) 48 performance indicators served to identify styles of play and can be used to assist with opposition analysis and in identifying a player's current style of play when analysed by the recruitment department; (2) the Dixon-Coles expected goals model has assisted in analysing a team’s quality which will assist opposition analysis and analysis of the current level a player is playing at; and finally; (3) teams can analyse what makes up the optimal style of play and adjust their style of play to improve results by copying other teams of a similar level who have better performances. |
|---|---|
| Item Description: |
Part of this thesis has been redacted to protect personal information |
| Keywords: |
Sports Science |
| College: |
Faculty of Science and Engineering |
| Funders: |
Swansea City Football Club |