Evaluating goal threat in football using player and ball locations

Abstract

Goal scoring in football is relatively low but vitally 
important, hence research has considered how goals are 
created and scored with measures such as expected goals 
prevalent. The dynamical systems theoretical perspective, 
considers a collective system, such as football, as existing 
in two states, stable (no substantive advantage for either 
team) or unstable (advantage present). Hence, goal scoring 
events occur when the system has become unstable, with a 
“perturbation” the event causing the system state change. 
Here, a “goal threat” value was calculated every second 
(scaled from 0 to 100) using the XY coordinates of players 
and the ball, weighted in relation to proximity to the goal 
(a potential proxy for the degree of system instability). 
Video recordings and synchronised Amisco 2D 
representations of goals (n=64) scored in Swansea City 
AFC English Premier League 2012/2013 matches (n=20) 
were analysed using Dartfish v10 Pro software. Each goal 
was analysed from when the play was judged to be stable 
(no obvious goal scoring opportunity), or the start of 
possession, until the goal had been scored. Goals were not 
always preceded by high goal threat values (maximum 
goal threat values ranged from 13.4 to 99.0). The authors 
independently subjectively determined that perturbations 
occurred up to 7 seconds from when the goal threat value 
increased by at least 40%. Thus, perturbations were not 
directly related to goal scoring opportunities. This novel 
method provides a useful, quantifiable, and simple 
measure of goal threat that may also aid audience 
engagement and measure defensive effectiveness