Event

The Department of Mathematics and Statistics will be hosting a 2nd Ph.D. Seminar presented by Mr. Brendon Bhagwandeen on the topic An Expected Goals (xG) Model and its Practical Applications in Football. The seminar will be held on Tuesday, June 20, from 9 - 10 a.m.

Zoom Link:   https://sta-uwi-edu.zoom.us/j/92236160843?pwd=clU3V1g3WjN0YThCTzdPdlczamcvZz09

Abstract:

The expected goal (xG) metric provides a more representative measure of team and player performance. The final result of a match often may not represent the true performance of teams and players. Therefore, it has become acceptable to use alternative statistics including shots on target, ball possession, chances created, and xG to gain a better understanding of player and team performance. To measure the probability of scoring a goal, several features are used to train an xG model based on event and tracking data from football matches. Selection of these features, the volume of data, and the algorithm affect the performance of any xG model. Data used in this process is usually imbalanced. We propose a hybrid resampling technique within an xG model using 361,307 shot-related events from eight seasons of football in the top 5 European leagues between 2014-15 and 2021-22. Aggregated team and player profiles are derived to obtain an interpretable xG model for evaluating both player and team performance.