The application of accelerometer-derived moving averages to quantify peak demands in basketball: A comparison of sample duration, playing role, and session type

The purpose of this was to study compare peak external workload intensities in basketball using accelerometer-derived moving averages between different sample durations (0.5–5 minutes), session types (training vs. game-play), and playing roles (starting vs. bench players). Five starting and 3 bench players were monitored over a 15-week competitive season using accelerometers. For all training sessions and games, peak external workload intensities were determined using accelerometer-derived moving averages for PlayerLoad per minute (PL·min-1) across sample durations of 0.5, 1, 2, 3, 4, and 5 minutes. Linear mixed-models and effect sizes (ESs) were used to compare peak PL·min-1 between sample durations, session type, and playing role. Peak PL·min-1 was significantly different between all sample durations (p < 0.05; ES = 0.88–5.45), with higher intensities evident across shorter sample durations. In starting players, peak intensities were significantly higher during games compared with training for all sample durations (p < 0.05; ES = 0.69–0.93). Peak game intensities were higher in starting players using all sample durations (p > 0.05; ES = 0.69–1.43) compared with bench players. Shorter sample durations produced higher peak PL·min-1. Peak intensities were higher during games than training in starting players, indicating training may not adequately prepare players for the most demanding passages of game-play.