Prediction of the penalty kick direction in men’s soccer

ABSTRACT The penalty kick might be a crucial moment during a soccer game because of its relevance on the scoreboard, mainly during shootouts. For this purpose, the present study established a model to predict the ball direction taking into consideration the leg dominance of the penalty taker and other posture variables just before the ball contact. A total of 412 penalties from 11 international high-standard tournaments (2010–2019) were analysed. The better predictive model for the shot direction was selected through all possible subsets regression method. Thus, the final model included three predictors and two multiplicative terms, with an acceptable classifying power with a specificity and sensitivity greater than 85%. This model was applied for predicting the probability to shot to the left and the right side of the goal in 12 different kicking patterns. The most frequent pattern was a right-footed taker with the left arm in a >90° abduction with a left orientation of the support foot (n = 122; 30%). Following this pattern, probability of shooting to the left or to the right was 0.970 (95% CI: 0.927–0.988) and 0.030 (95% CI: 0.012–0.073), respectively. The present study conceived a model as parsimonious as possible with three predictors easily available to the goalkeeper.