The aims was to model a latent growth curve of young swimmers' performance over a season and its relationship with biomechanics, quantify the direct effect of the exogenous variables selected on performance across a season and analyze the sex effect on the performance growth. Fourteen boys (12.33±0.65 years) and 16 girls (11.15±0.55 years) were evaluated. Performance, stroke frequency, speed fluctuation, arm's propelling efficiency, active drag, active drag coefficient and power to overcome drag were collected in four different moments of the season. Latent growth curve modeling was computed to understand the longitudinal variation of performance (endogenous variables) over the season according to the biomechanics (exogenous variables). Latent growth curve modeling showed a high inter- and intra-subject variability in the performance growth. Sex had a significant effect at the baseline and during the performance growth. In each evaluation moment, different variables had a meaningful effect on performance (M1: Da, β=-0.62; M2: Da, β=-0.53; M3: ηp, β=0.59; M4: SF, β=-0.57; all P<0.001). The models' good-of-fit was 1.40≤x2/df≤3.74 (good-reasonable). Latent modeling is a comprehensive way to gather insight about young swimmers' performance over time. Different variables were the main responsible for the performance improvement. A gender gap, intra- and inter-subject variability was verified.
