Analyzing Soccer Training Sessions to Predict Player Match Performance
Poster Number
Withdrawn
Introduction/Abstract
This research analyzes performance data from training sessions of male soccer players to predict player performance in subsequent matches. As a result, this research provides a prediction analysis and machine learning model for player match performance. The model is developed based on training performance in sessions leading up to games as well as the resulting performance data in various matches. This is then used on current training data in order to predict each player's match performance capabilities prior to their next match taking place.
Purpose
There are two overarching purposes of this research and reasons regarding its importance. The first is to identify how various training loads leading up to a match affect player outcomes. The second is to create a platform for coaches to optimize their starting lineup and player substitutions throughout the duration of the game based on performance predictions.
Method
Specifically, this research pertains to Division 1 male soccer players attending University of the Pacific. The data analyzed is gathered from games and sessions from both the Fall 2021 Season and Spring 2022 Off-Season. Each training session consists of a series of activities, all of which have varying intensities that differ from the intensity level experienced on a game day. In order to account for this, a unitless coefficient is applied to the various attributes allowing for the training and match data to be directly comparable. The main attributes being considered fall into two categories: volume and intensity. Volume is measured by distance, sprint distance, power plays, energy, sprints, and impacts. While intensity is measured by top speed, mean distance, power score, work ratio, and player load. All of these contribute to the athletic performance of each individual player and can be compiled into an overarching session score relative to that performance.
Results
The hope for this research is that these results will be utilized to guide and optimize match preparation in the future.
Significance
Both of the purposes mentioned earlier are directed towards aiding the coaching staff in optimizing the success and performance of both the players as individuals and as a team. By identifying and properly applying this information, teams will experience a significant advantage against their opponents.
Location
William Knox Holt Memorial Library and Learning Center, University of the Pacific, 3601 Pacific Ave., Stockton, CA 95211
Format
Poster Presentation
Poster Session
Afternoon
Analyzing Soccer Training Sessions to Predict Player Match Performance
William Knox Holt Memorial Library and Learning Center, University of the Pacific, 3601 Pacific Ave., Stockton, CA 95211
This research analyzes performance data from training sessions of male soccer players to predict player performance in subsequent matches. As a result, this research provides a prediction analysis and machine learning model for player match performance. The model is developed based on training performance in sessions leading up to games as well as the resulting performance data in various matches. This is then used on current training data in order to predict each player's match performance capabilities prior to their next match taking place.
