Deep soccer analytics: learning an action-value function for evaluating soccer players

We develop a new approach to evaluating all types of soccer actions from play-by-play event data

Guiliang Liu; Yudong Luo; Oliver Schulte; Tarak Kharrat

2020

Scholarcy highlights

  • The classifier is implemented with a neural network rather than CatBoost in due to the size of dataset
  • We are indebted for helpful discussion and comments to Norm Ferns, Evin Keane, and Bahar Pourbabee from Sportlogiq
  • We denote the number of times such a transition occurs as
  • Where the \('\) indicates the successor triple. We freely use this notation for marginal counts as well, for instance
  • Line uses the empirical estimate of the expected Q-value \(Q_{{ team}}(s',a')]\) given that player i acts computed from the maximum likelihood estimates of the transition probabilities:

Need more features? Save interactive summary cards to your Scholarcy Library.