The development of talent may take different forms for youth athletes, but there are only few longitudinal studies tracking this process. This study prospectively examines: 1) the development of talented Dutch tennis players aged 11.5 to 16, and 2) the predictability of performance at age 16, based on players’ performance in preceding years. Biannually, we collected performance rating data of 1010 players (645 males, 365 females), who were ranked top-300 in the national youth ranking at least once. Applying Growth Mixture Modelling, we found that a 4-class model fit the male sample, while a 5-class model fit the female sample. The best male class displayed non-linear development, improving slowly until age 13.5, after which development accelerated. The best female class showed consistent linear improvement. The correlation between performance at age 16 and preceding ages was consistently higher for the female sample. Additionally, the overlap percentage between the 5% best players at age 16 and preceding ages was consistently higher for the female sample as well. Our findings shed a new light on male-female differences in talent development trajectories, and suggest that early talent identification could be more feasible for female players compared to male players.

Files: Scripts for the analyses in study titled: "Developmental performance trajectories of talented tennis players during adolescence: a prospective longitudinal approach". Preprocessing starts in Python scipt ("Rating_analysis_python.py"), then the GMM analysis is performed using the R script("Rating_analysis_R.R"), finally the additional analyses are finished in the Python script("Rating_analysis_python.py").