Academic Journal - Journal of Big Data - 2025
Jiangyan Yang, Huanmin Ge and Yixiong Cui
https://journalofbigdata.springeropen.com/counter/pdf/10.1186/s40537-025-01128-3.pdf
The aim of this journal was to develop an artificial intelligence framework that can automatically detect counterattack sequences in football matches. This model will use deep learning to analyse how players make decisions such as such as passes, dribbles or shots during counterattacks. It will help assess whether certain players actions were effective and what spatial/temporal features (e.g. player positions, distances, angles) influenced the outcomes.
This model mainly uses event data such as passes, shot or ball recoveries combined with tracking data for complete premier league matches. These datasets are collected and processed after matches not in real time during the match.
The purpose of this was to detect counterattack sequences, evaluate decision making qualities and derive tactical insights such as how often teams counterattack, who’s initiates then and what spacial patterns lead to success. These insights can be used for tactical review post match, coaching feedback and performance analysis
While it is not currently implemented, the author of this highlighted the need for live analysis as it is built on automated detection of counterattack sequences and if tracking data were streamed in real time, it could flag emerging counterattacks live. This could support alerting analysts when a team is vulnerable to counter.