Space evaluation in football games via field weighting based on tracking data

Takuma Narizuka*, Yoshihiro Yamazaki, Kenta Takizawa

*Corresponding author for this work

Research output: Contribution to journalArticlepeer-review

1 Citation (Scopus)

Abstract

In football game analysis, space evaluation is an important issue because it is directly related to the quality of ball passing or player formations. Previous studies have primarily focused on a field division approach wherein a field is divided into dominant regions in which a certain player can arrive prior to any other players. However, the field division approach is oversimplified because all locations within a region are regarded as uniform herein. The objective of the current study is to propose a fundamental framework for space evaluation based on field weighting. In particular, we employed the motion model and calculated a minimum arrival time τ for each player to all locations on the football field. Our main contribution is that two variables τof and τdf corresponding to the minimum arrival time for offense and defense teams are considered; using τof and τdf, new orthogonal variables z1 and z2 are defined. In particular, based on real datasets comprising of data from 45 football games of the J1 League in 2018, we provide a detailed characterization of z1 and z2 in terms of ball passing. By using our method, we found that z1(x→ , t) and z2(x→ , t) represent the degree of safety for a pass made to x→ at t and degree of sparsity of x→ at t, respectively; the success probability of passes could be well-fitted using a sigmoid function. Moreover, a new type of field division approach and evaluation of ball passing just before shots using real game data are discussed.

Original languageEnglish
Article number5509
JournalScientific reports
Volume11
Issue number1
DOIs
Publication statusPublished - 2021 Dec

ASJC Scopus subject areas

  • General

Fingerprint

Dive into the research topics of 'Space evaluation in football games via field weighting based on tracking data'. Together they form a unique fingerprint.

Cite this