Prescient Perspectives on Football Tactics: A Case with Liverpool FC, Corners and AI

Prescient (not prescientific) ideas are those that challenge the assumptions that predominate a given field and that typify beforehand the future of that domain. In this short communication, we question how much we are willing to believe that predictive Artificial Intelligence is already able to suggest good prescient ideas on football, for example by hinting the tactical adjustments needed to shoot corner kicks that finally end in goals. We have conducted statistical Chi-squared tests that show that the Liverpool Football Club, in the Premier League seasons 2021/22, 2022/23 and 2023/24, has reached a higher potential for goals from corners with respect to the previous seasons 2016/17, 2017/18 and 2018/19, only in the attacking phase of play (scored goals), but not in the defensive one (conceded goals). This despite a multi-year collaboration between Liverpool FC and authors of [1] on AI, reportedly begun during the season 2020/21. Moreover, no statistically significant difference was found, with another series of Chi-squared tests, between the goals scored/conceded from corner kicks by Liverpool FC and those scored/conceded, in the seasons 2021–24, by Arsenal FC and Manchester City FC. Our study suggests preliminary evidence against the belief that AI is ready to prompt prescient ideas in a dynamic domain, characterized by many multi-agent interactions, like football. Limitations to these results come from the uncertainty on the real nature of the collaboration between Liverpool FC and AI on corner kicks, as discussed at length in the paper