Field-based Fitness Assessments Have Limited Predictive Insight Into Game Performance In High School Basketball Athletes

Purpose: Field-based fitness tests are frequently used to assure athlete readiness for competition. However, there is limited research on the effectiveness of fitness tests in predicting competition performance in high school basketball athletes. Our aim was to understand if common field-based fitness assessments used by high school sports academies, could effectively predict performance and which variables are most valuable. METHODS: Data from 239 youth basketball athletes [M/F: 202/36; 16.1 ± 1.6 yrs; 183.0 ± 11.3 cm; 76.8 ± 14.4 kgs] were collected during the 2022-2023 and 2023-2024 school years were utilized in this analysis. Predictors included variables from the YoYo-intermittent Test (YoYo), 1-mile run time (1MR), 5-10-5 pro agility (PA), Nordboard Hamstring Strength (NB), Isometric Midthigh Pull (IMTP), and Countermovement Vertical Jump (VJ). To determine game performance Hudl Game Stats were utilized (including value points system (VPS), points, rebounds, blocks, and turnovers). VPS is considered an overall performance indicator that is comprised of several in games statistics. To identify the optimal prediction model for each outcome a partial least squares regression (PLS) was used identify possible predictor variables. The PLS was followed by a stepwise regression to further identify associated predictors. RESULTS: No model included multiple significant predictors. IMTP peak power was the strongest predictor for rebounds, blocks, and turnovers (R2 = 16.6%, 12.7%, 4.9%, respectively; p’s < 0.006). Peak takeoff force during the VJ was the strongest predictor of points scored (R2 = 10.3%; p < 0.001). Lastly, 1MR time was the strongest inverse predictor for VPS (R2 = 6.9%; p = 0.002). CONCLUSION: The isometric mid-thigh pull, vertical jump, and one-mile run were statistically significant, albeit relatively weak predictors of basketball game performance in high school basketball athletes, with no metric predicting more than 17% of the variance in-game performance outcomes. Future research should integrate non-fitness related covariates that may provide additional predictive insight such as, anthropometrics, age, and practice-specific performance. Supported by: No grant funding was obtained for this investigation