Will Positional Shifts Replace Formations? A Tactical Revolution in Football!

The representative dataset includes raw positional (x, y) coordinates for all players during goal-scoring sequences, structured across six pre-goal time intervals. The accompanying preprocessing scripts implement spatial normalization (0–1 scaling), Euclidean distance calculations, similarity ratio matrices (1 − normalized distance), fixed-effects panel regression modelling, and supervised machine learning classification (Random Forest and XGBoost) based exclusively on positional configurations. Together, these materials allow full replication of the analytical workflow described in the manuscript.