Modelling the Collective Movement of Football Players

Since the development and rise of different methods for obtaining tracking data from sports games, data analysis has played a very important role in the world of football. With many studies and models having been created for carrying out a posteriori analyses of football games, there is a necessity to take a step forward and include collective motion simulations in these analyses. With this thesis, a new model for computing the optimal positions of football players during an attacking situation is proposed. Three different computational models that quantify paramount aspects of football such as pitch control, pass success probabilities and pass impact will be used as the basis of our simulations. The new developed model built on top of these is able to look forward one or two seconds into the future and predict the optimal positions of the attacking players according to certain simple rules that are set to maximise or minimise different combinations of the mentioned algorithms