Repository logo
 

Analysing Opponent Positioning Relative to the Ball: A Predictive Model for Football Possession Outcomes

Files

Dissertation(1.04 MB)

Date

2024

Authors

Supervisor

Spencer, Kirsten

Item type

Dissertation

Degree name

Master of Sport, Exercise and Health

Journal Title

Journal ISSN

Volume Title

Publisher

Auckland University of Technology

Abstract

This dissertation investigates the factors influencing possession outcomes in football using quantitative research methodologies, specifically multinomial logistic regression. Although there is a considerable amount of literature that presents regression models to predict outcomes in football matches, none investigated opposition player positioning relative to the ball. The research aims to provide an objective measure of team performance by analysing various contextual and situational variables, in order to accurately predict possession outcomes. The study focuses on the Wellington Phoenix Academy Men’s Reserve Team during their 2023 Central League campaign. The research aims to provide an objective measure of team performance by analysing various contextual and situational variables. Routinely collected data, through foot-mounted inertial measurement units, was provided by the Wellington Phoenix Academy. A total of 923 ball possessions from a 15 game sample were analysed using a multinomial logistic regression model. The findings suggested the positioning of the opponents relative to the ball does have an impact on the chances of creating successful possession outcomes. As the number of opponents behind the ball at the end of a possession increased, the probability of a successful outcome increases, when comparison against an unsuccessful outcome. Progressive passes and backwards passes were positively correlated and significantly predicted successful outcomes. Situational variables were analysed (adapted from literature on professional matches) however these were not effective predictors. This study presented a novel way in which to value actions, and the findings indicated that this model is a sufficient predictor tool. Ultimately, these insights contribute to a deeper understanding of tactical decision-making and performance analysis in football, emphasizing the utility of data-driven approaches in enhancing coaching practices. Coaching staff could implement this model to assess what actions in possession are most effective for creating goal scoring opportunities.

Description

Keywords

Source

DOI

Publisher's version

Rights statement

Permanent link

http://hdl.handle.net/10292/18235

Collections

Masters Dissertations