Integrating video tracking and GPS to quantify accelerations and decelerations in elite soccer

Authors:
Open Access
Publisher website
Google Scholar
Abstract
The aim of this study was to analyze the degree of agreement comparing number and distance covered in different acceleration and deceleration sections registered by a video tracking system (MEDIACOACH) and a GPS device (WIMU PRO) during official competition. Data from a Spanish professional club were registered over the course of a season. First, the descriptive statistics presented more bursts of accelerations and decelerations in WIMU PRO than in MEDIACOACH, whereas the distances covered recorded by both systems were similar. Second, negative relationships were found (i.e., negative bias) comparing WIMU PRO to MEDIACOACH in the number of accelerations and decelerations between 0/1 m/s2 and ½ m/s2 (p < 0.05), and in the distances covered in accelerations and decelerations (p < 0.05) between 0/1 m/s2 and in accelerations and decelerations registered between 2/3 m/s2 and more than 3 m/s2. Moreover, the differences in means (i.e., standardized mean bias) across the two devices were trivial (> 0.19) and small (0.2–0.59) for most variables. The standardized typical errors in the estimate (TEE) were moderate (0.3–0.59) and small to moderate (0.1–0.29 to 0.3–0.59), respectively. Also, the Intra class Correlation Coefficients (ICCs) for agreement and consistency between systems showed good and excellent values (> 0.90). The magnitude of change in means (%) between systems, defined as the percentage change between the numbers or values, was below 14% and 7% for number and distances covered, respectively. All scores in the smallest worthwhile change were lower than 9% and in the coefficients of variation were lower than 95% and 15%, respectively. Thus, both systems demonstrated an acceptable degree of agreement and could be useful in analyzing players’ acceleration demands in professional soccer. However, caution is required when interpreting the results and a comparison with a gold standard is required in order to validate both systems.

This publication has 48 references indexed in Scilit: