Perspectives on data analytics for gaining a competitive advantage in football: computational approaches to tactics

The role of data-driven analyses is becoming more prominent in football. These have the potential to impact decision-making processes for team performance and player recruitment. Research in this area makes use of large datasets consisting of event and tracking data from multiple teams, leagues and seasons. A well-known computational solution is the Expected Goal model for post-match analysis and operational decision-making.Despite a shared research interest in football tactics, computational research in football is somewhat disconnected from the sports science community. We believe that there is much to gain from a closer collaboration between these disparate communities. To this end, the present commentary has three goals. First, we want to synthesize the historical computational work in areas such as evaluating tactics, predicting player and team success, and modeling players' movements. This work has largely been published in technical computational venues, and hence we hope to provide an access point for those interested in learning about this area. Second, we will highlight some emerging topics, such as automating parts of match analysis and analyzing decision-making. These are topics that require an in-depth collaboration with domain experts and therefore would benefit from a tighter integration among these communities. Third, we would like to discuss some advice and initiatives that we hope will be helpful in strengthening the ties between these communities.