Selecting Training-Load Measures to Explain Variability in Football Training Games

The purpose of this study was to investigate the structure of interrelationships among external (eT) and internal (iT) training intensity metrics and how these vary depending on game format in soccer. The variables were collected from 16 semi-professional players in seven types of small, medium, large-sided, and simulated games (SG). The eT variables were (per min): peak velocity (Vmax), total distance (DTmin), distance covered at velocities less than 60% (D < 60%min), between 60 and 80% (D > 60%min), and more than 80% (D > 80%min) of the maximal velocity, player load (PLmin), and distance covered accelerating at more than 2 m⋅s-2 (Daccmin) and decelerating at less than −2 m⋅s-2 (Ddecmin). The iT variables were: Edwards arbitrary units (EDWmin) and time spent at more than 80% of the maximal heart rate (T > 80% HRmin). All game formats were represented by three principal components (PC), explaining from 66.9 to 76.0% of the variance. The structure of the interrelationships among variables involved similar distributions in the PCs that are related to energetic production systems, such as the strength/neuromuscular dimension (PLmin and/or Daccmin and Ddecmin, complemented by DTmin and D < 60%min), the endurance/cardiovascular dimension (EDWmin), and the velocity/locomotion dimension (Vmax, D > 60%min, or D > 80%min). A particular combination of external and internal intensity measures is required to describe the training load of game formats.