Baseline Performance and the 72-Hour Performance Window Following Plyometric Withdrawal in Collegiate Sprinters

Plyometric studies emphasise pre-post change, but coaches need to know what remains after the block ends. This single-cohort study followed 26 collegiate sprinters through 12 weeks of Monday plyometric training and 4 weeks of withdrawal, without a parallel control group. During intervention, 30 m sprint time and arm-assisted countermovement jump (CMJ) height were measured at 0, 24, 48, 72, and 96 h; 0 and 72 h were retained during withdrawal. Mixed models showed large week effects for sprint, F(1, 1534) = 2272.01, partial η² = 0.597, 95% CI [0.570, 0.622], and CMJ, F(1, 1534) = 2537.82, partial η² = 0.623, 95% CI [0.598, 0.647]. Time-point effects were evident, but week × time-point interactions were not. Relative to 0 h, sprint was 0.0098 s slower at 24 h and 0.0057 s faster at 72 h; CMJ was 0.311 cm lower at 24 h and 0.245 cm higher at 72 h (all Tukey-adjusted p < 0.001). Linear, quadratic, and autoregressive integrated moving average (ARIMA) models were compared by rolling one-step error. ARIMA had the lowest rolling errors and provided exploratory, model-based references for weeks 13-16. During withdrawal, 1/4 sprint and 0/4 CMJ 0 h observations fell within the 95% prediction interval; mean absolute deviations were 0.0233 s and 0.794 cm. Three of four 72 - 0 h observations were within the interval for both outcomes. The 72 - 0 h series more often fell within their model-based intervals than the 0 h series, indicating different short-term patterns.