Introduction: Talent detection is a dynamic and multifactorial process that must start at school. Objective: Create a mathematical model for evaluating the sporting potential of schoolchildren for athletics in speed, throwing, and endurance events and to test its psychometric properties. Methods: 2871 schoolchildren of both sexes, from 11 to 17 years old, from a military school participated. Between 2015 and 2019, students were submitted to a multidimensional battery of tests containing anthropometric, physical-motor, psychological, socio-environmental, maturational, and performance indicators. In addition, ten teachers evaluated the students regarding the intangibles aspects of their sporting potential and the expectation of future success during this period. Adopting analytical and heuristic procedures, the Gold Score Athletics was created - linear, hybrid (tests + coaches´ eye), and weighted index, according to each indicator's importance, depending on the event type. Results: In the model validation sample (n = 1384), 13.9%, 16.6%, and 11.7% of boys and 10.9%, 10.1%, and 9.1% of girls were classified as high potential (Gold Score = 60) for speed, throwing and endurance events, respectively. Internal consistency (r = 0.76 to 0.82) and diagnostic stability were high (r = 0.72 to 0.81). The Gold Score Athletics for sprinters, throwers, and long-distance runners, both for boys and girls, was higher in students selected for a national competition when compared to those not selected (p < 0.001; d: 0.95 a 1.44) - construct validity - and higher in medalists in an athletics competition, held two years after diagnosis, when compared to non-medalists (p < 0.05; d: 0.62 a 1.87) - predictive validity. Conclusion: The Gold Score Athletics is a valid and reliable scientific model for evaluating the sport's potential of schoolchildren, being useful in the talents detection for Athletics.
© Copyright 2023 Revista Brasileira de Medicina do Esporte. Sociedade Brasileira de Medicina do Esporte. Alle Rechte vorbehalten. / Übersetzt mit https://www.DeepL.com/Translator