A combined logistic regression and decision tree analysis of factors influencing throwing velocity and accuracy in little league baseball players

Background We investigated factors related to pitching performance—defined by pitch velocity and accuracy—based on baseball experience, pitching workload, anthropometrics, and range of motion and strength (grip and shoulder) among little league baseball players. Methods This cross-sectional study included 348 right-handed baseball players aged 9–12 years. Assessments included baseball and pitching experience, weekly pitching workload during both games and practices, anthropometrics, range of motion, grip, and shoulder strength. Throwing velocity was measured using a radar gun during a standardized flat-ground throwing task, and throwing accuracy was assessed using a nine-zone target scoring system. Players were classified into high- and low-velocity groups based on the median velocity to enable decision tree analysis. Using logistic regression and decision tree analyses, factors associated with throwing velocity and accuracy were identified. Results Logistic regression analysis revealed that pitcher experience (odds ratio [OR]:3.45; P = .003), pitching workload ≥100/week (OR: 5.16; P = .002), and dominant-side grip strength (OR: 1.17; P = .033) were significantly associated with high pitch velocity. Decision tree analysis showed that players with a strong dominant-side grip (>14.5 kgf), age >10 years, and grip strength >19.8 kgf had the highest probability of high pitch velocity. Among players aged ≤10 years, those with ≥3 years of experience were more likely to have high velocity. Only baseball experience was a significant factor associated with pitch accuracy (OR: 1.23; P = .040). High-accuracy players were taller, older, and had greater nondominant grip strength than low-accuracy players. Conclusion Throwing velocity was modestly associated with pitching experience, weekly pitching workload, and dominant-side grip strength, whereas throwing accuracy was associated primarily with baseball experience. These findings should be interpreted cautiously given the modest effect sizes and cross-sectional design.