The Clustering Project

A new framework for smarter scouting

• Written by Soccerment Research and Antonio Gagliardi (Assistant Coach & Head of Match Analysis at Italy National Team FIGC)
• Foreword by Fabio Paratici (Managing Direct, Football - Tottenham Hotspurs)
• 160 pages

Abstract

Using detailed stats from season 2017/18 to 2020/21, we applied a clustering analysis to categorise players into thirteen groups, based on their function on the pitch. We structured our analysis to reconstruct not only a class label for each player, but a full hybridation profile, with the classification probability for all thirteen clusters, allowing us to recover a more unique (albeit not conclusive) characterisation.

These results have an immediate application in scouting, providing a search filter with a data-driven characterisation, as opposed to the a priori one given by traditional roles.