This study describes an approach to quantification of attacking performance in football. Our procedure determines a quantitative representation of the probability of a goal being scored for every point in time at which a player is in possession of the ball–we refer to this as dangerousity. The calculation is based on the spatial constellation of the player and the ball, and comprises four components: (1) Zone describes the danger of a goal being scored from the position of the player on the ball, (2) Control stands for the extent to which the player can implement his tactical intention on the basis of the ball dynamics, (3) Pressure represents the possibility that the defending team prevent the player from completing an action with the ball and (4) Density is the chance of being able to defend the ball after the action. Other metrics can be derived from dangerousity by means of which questions relating to analysis of the play can be answered. Action Value represents the extent to which the player can make a situation more dangerous through his possession of the ball. Performance quantifies the number and quality of the attacks by a team over a period of time, while Dominance describes the difference in performance between teams. The evaluation uses the correlation between probability of winning the match (derived from betting odds) and performance indicators, and indicates that among Goal difference (r = .55), difference in Shots on Goal (r = .58), difference in Passing Accuracy (r = .56), Tackling Rate (r = .24) Ball Possession (r = .71) and Dominance (r = .82), the latter makes the largest contribution to explaining the skill of teams. We use these metrics to analyse individual actions in a match, to describe passages of play, and to characterise the performance and efficiency of teams over the season. For future studies, they provide a criterion that does not depend on chance or results to investigate the influence of central events in a match, various playing systems or tactical group concepts on success.

An important structuring feature of a soccer match is the in-game status, whether a match is interrupted or in play. This is necessary to calculate performance indicators relative to the effective playing time or to find standard situations, ball actions, and other tactical structures in spatiotemporal data. Our study explores the extent to which the in-game status can be determined using time-continuous player positions. Therefore, to determine the in-game status we tested four established machine learning methods: logistic regression, decision trees, random forests, and AdaBoost. The models were trained and evaluated using spatiotemporal data and manually annotated in-game status of 102 matches in the German Bundesliga. Results show up to 92% accuracy in predicting the in-game status in previously unknown matches on frame level. The best performing method, AdaBoost, shows 81% precision for detecting stoppages (longer than 2 s). The absolute time shift error at the start was ≤ 2 s for 77% and 81% at the end for all correctly predicted stoppages. The mean error of the in-game total distance covered per player per match using the AdaBoost in-game status prediction was − 102 ± 273 m, which is 1.3% of the mean value of this performance indicator (7939 m). Conclusively, the prediction quality of our model is high enough to provide merit for performance diagnostics when teams have access to player positions (e.g., from GPS/LPM systems) but no human-annotated in-game status and/or ball position data, such as in amateur or youth soccer.