Contextualizing Physical Data in Professional Handball: Using Local Positioning Systems to Automatically Define Defensive Organizations

Abstract

In handball, the way the team organizes itself in defense can greatly impact the player’s activity and displacement during the play, therefore impacting the match demands. This paper aims (1) to develop an automatic tool to detect and classify the defensive organization of the team based on the local positioning system data and check its classification quality, and (2) to quantify the match demands per defensive organization, i.e., defining a somehow cost of specific defensive organizations. For this study, LPS positional data (X and Y location) of players from a team in the Spanish League were analyzed during 25 games. The algorithm quantified the physical demands of the game (distance stand, walk, jog, run and sprint) broken down by player role and by specific defensive organizations, which were automatically detected from the raw data. Results show that the different attacking and defending phases of a game can be automatically detected with high accuracy, the defensive organization can be classified between 1–5, 0–6, 2–4, and 3–3. Interestingly, due to the highly adaptive nature of handball, differences were found between what was the intended defensive organization at a start of a phase and the actual organization that can be observed during the full defensive phase, which consequently impacts the physical demands of the game. From there, quantifying for each player role the cost of each specific defensive organization is the first step into optimizing the use of the players in the team and their recovery time, but also at the team level, it allows to balance the cost (i.e., physical demand) and the benefit (i.e., the outcome of the defensive phase) of each type of defensive organization.

Analysis of Movement and Actions of Wingers as Second-Line Players in Organized Attack in Handball

“…The correspondence analysis results also showed that the total of attacks is divided into two groups (vertical axis): a) attacks of long duration with a winger running in as a second-line player, and b) attacks of short duration with a winger running in as a second-line player. This makes sense because of the different evolution and the final outcome of the attack, due to the fact that attacks are temporarily interrupted by faults (long duration) or completed in a very short time (short duration) ( Guignard et al, 2022 ).…”

“…C) Attacks of long duration (more than ten seconds) (top right in Figure 1 ). This is because defenses adapt and try to make faults and interrupt the evolution of the attack early ( Guignard et al, 2022 ). Finally, D) Attacks where the winger enters as a second-line player, “slides” and stays without taking the ball (bottom right in Figure 1 ).…”

A deep dive into the use of local positioning system in professional handball: Automatic detection of players’ orientation, position and game phases to analyse specific physical demands

“…First, an automated phase recognition algorithm was designed to detect game phases through positional data in handball. In the same idea, automation of defensive patterns was already developed by Guignard et al [ 13 ]. When the phase recognition algorithm was applied to the validation sets, there is an average of 98.95% of the time where the phase was correctly recognized.…”

“…Data come from the same sample used in Guignard et al [ 13 ]. We collected the data from the WIMU PRO system (RealTrack Systems S.L., Almería, Spain) with sensors placed on the back of the athletes between the two scapulas at the level of the C7 vertebra.…”

“…of 3rd order) to have every variables on the same timeline and avoid synchronisation issues. With the same procedure than Guignard et al [ 13 ], the positioning reference, initially defined outside the boundaries of the court, has been relocated to the centre of the court (on the median lines on both X and Y axes). by using the barycentre of the positions of the two GKs when they were in front of their goal.…”