Velocity zone classification in elite women’s football: where do we draw the lines?

ABSTRACT

Objectives: This study aims to develop generic velocity thresholds for the analysis of external load data collected in international women’s football matches.

Methods: Doppler-derived recordings of instantaneous velocity and acceleration were collected (10 Hz GPS) from 27 international female football players during 52 international matches between 2012 and 2015. Data were examined with k-means, Gaussian mixture model (GMM), and Spectral Clustering methods to identify four velocity zones, in each completed half of match-play (277 observations). Spectral Clustering was also performed with 4 different smoothing parameters ($\beta$ values of 0, 0.001, 0.01, and 0.1). Linear-mixed modelling was used to determine generic squad thresholds, accounting for the within-subject variation.

Results: k-means and GMM generated low transition velocities, which had limited logical validity and deemed not fit for purpose. Spectral Clustering with a $\beta$ value of 0.1 derived thresholds that differed from the various methods adopted in existing literature and industry practice, yet providing a rigorous, acceptable, and feasible determination of velocity thresholds.

Conclusion: Velocities of 3.46 (12.5 km h⁻¹), 5.29 (19.0 km h⁻¹), and 6.26 m s⁻¹ (22.5 km h⁻¹) are recommended as entry criteria into high, very-high velocity, and sprinting locomotor categories, respectively, for the purpose of external load assessments in elite women’s football.

KEYWORDS:

• GPS
• football
• velocity thresholds
• data-mining
• female

Disclosure statement

No potential conflict of interest was reported by the authors.

Supplementary Material

Supplementary can be accessed here.