Expected Possession Value of Control and Duel Actions for Soccer Player's Skills Estimation

TL;DR

This work extends the Expected Possession Value ($EPV$) framework for soccer by introducing a decay-based weighting, improved possession risk via effective playing time, and explicit modeling of aerial and ground duels. It develops six $EPV$ models that integrate $xG$ context, duel outcomes, and player skills to compute season-long rewards (PCR) and to predict PCR for the upcoming season, while addressing data biases such as transfers and presence-only data. The methodology includes a modified $Glicko ext{-}2$ rating for duel contexts, specialized $xG$ models for open-play and set-pieces, and a robust feature set (~600 features) that accounts for league and team strength. The approach aims to improve player selection and performance forecasting in diverse leagues by better capturing the dynamics of possession and duels, with potential extensions to injury data and more sophisticated transfer-context modeling.

Abstract

Estimation of football players' skills is one of the key tasks in sports analytics. This paper introduces multiple extensions to a widely used model, expected possession value (EPV), to address some key challenges such as selection problem. First, we assign greater weights to events occurring immediately prior to the shot rather than those preceding them (decay effect). Second, our model incorporates possession risk more accurately by considering the decay effect and effective playing time. Third, we integrate the assessment of individual player ability to win aerial and ground duels. Using the extended EPV model, we predict this metric for various football players for the upcoming season, particularly taking into account the strength of their opponents.