Individual ball possession in soccer

Abstract

This paper describes models for detecting individual and team ball possession in soccer based on position data. The types of ball possession are classified as Individual Ball Possession (IBC), Individual Ball Action (IBA), Individual Ball Control (IBC), Team Ball Possession (TBP), Team Ball Control (TBC) und Team Playmaking (TPM) according to different starting points and endpoints and the type of ball control involved. The machine learning approach used is able to determine how long the ball spends in the sphere of influence of a player based on the distance between the players and the ball together with their direction of motion, speed and the acceleration of the ball. The degree of ball control exhibited during this phase is classified based on the spatio-temporal configuration of the player controlling the ball, the ball itself and opposing players using a Bayesian network. The evaluation and application of this approach uses data from 60 matches in the German Bundesliga season of 2013/14, including 69,667 IBA intervals. The identification rate was F = .88 for IBA and F = .83 for IBP, and the classification rate for IBC was κ = .67. Match analysis showed the following mean values per match: TBP 56:04 ± 5:12 min, TPM 50:01 ± 7:05 min and TBC 17:49 ± 8:13 min. There were 836 ± 424 IBC intervals per match and their number was significantly reduced by -5.1% from the 1st to 2nd half. The analysis of ball possession at the player level indicates shortest accumulated IBC times for the central forwards (0:49 ± 0:43 min) and the longest for goalkeepers (1:38 ± 0:58 min), central defenders (1:38 ± 1:09 min) and central midfielders (1:27 ± 1:08 min). The results could improve performance analysis in soccer, help to detect match events automatically, and allow discernment of higher value tactical structures, which is based on individual ball possession.

Fig 1. Hierarchy of soccer constructs.

Individual Ball Possession and its subtypes are a fundamental prerequisite for being able to discern ball actions. These actions are again the base for identifying situations. The typical behavior in these situations can be aggregated to playing styles.

Fig 2. Relationships between the different types of ball possession.

Team ball possession (TBP) consists of an unbroken sequence of individual ball possession (IBP) phases. IBP is composed of one or more phases in which an action can be performed with the ball. These are designated individual ball actions (IBAs). It is possible for IBAs to be separated by void phases during which no ball control exists (e.g. while the ball is in the air). An IBC is an IBA where ball control is present. Team ball control (TBC) is the union of all TBC phases. Team play making (TPM) corresponds to TBC plus the intermediate void phases.

Fig 3. Logical procedural steps for detecting different types of ball possession.

Detection is based on identifying the starting and endpoints of IBP and IBA.

Fig 4. Illustration of IBA starting point determination according to the naive physics model, ball possession exists at time t when the separation d between player(x_ti) and Ball (b_t is less than the threshold T_p.

Ball possession is attributed to the player closest to the ball (smallest ${\underset{\_}{x}}_t^1$ here). With kick detection (right), the acceleration of the ball ${\left|\right|\underset{\_}{\ddot{b}}}_t\left|\right|$ (smoothed, norm of xy-components) must simultaneously show a local maximum of at least 4 ms^-2.

Fig 5. Detection of IBA endpoints: An endpoint exists as soon as the player is unable to interact with the ball for a period of one second.

This is the case after a shot on goal or a pass, for example, but not when the player is dribbling the ball stated formally, when using the ball prediction method, the endpoint t of an IBA phase is arrived at when the distance d between ball position ($b_t^{+}$) and player position (${\underset{\_}{x}}_t^i$) is no longer below the threshold value (T_A) for a period of one second (t⁺) at the ball's current speed and direction (${\underset{\_}{\dot{b}}}_t$). The player's position is assumed to be constant. The player prediction method additionally takes into account the player's future position $x_t^{i+}$ based on his direction and speed (${\underset{\_}{\dot{x}}}_t^i$).

Fig 6. Heatmaps based on all positions (left) compared to positions during IBC (right).