Optimizing Pre-Competition Taper Strategies in Speed Climbing Athletes Using Session-RPE and Countermovement Jump: A Banister Impulse-Response Model Approach

This study aimed to develop a performance prediction model for speed climbing based on session-RPE (sRPE) and countermovement jump (CMJ), and to determine optimal pre-competition taper strategies. Training data were collected from 12 speed climbers over approximately 180 days (931 daily observations), including daily sRPE, CMJ, and climbing performance. The Banister fitness–fatigue impulse-response model was employed to quantify the dose–response relationship of training load, combined with random forest and gradient boosting for nonlinear feature analysis. Parametric grid search was used to optimize taper duration and daily load sequences. CMJ emerged as the most stable predictor of climbing performance (Pearson r = −0.319), ranking first in random forest feature importance (23.4%), while same-day sRPE showed negligible correlation with performance (r = 0.043). Group-level Banister model estimates showed a fitness time constant of τ1 = 34 ± 19 days and a fatigue time constant of τ2 = 17 ± 5 days, yielding τ1/τ2 = 2.0. Optimal taper duration ranged from 5 to 21 days, with load reduced to 29–90% of normal training load, exhibiting highly individualized characteristics. The physiological basis of tapering lies in exploiting the time gap τ1 > τ2—allowing fatigue to dissipate before fitness decays. A taper window of τ2 to τ1 days (approximately 17–34 days) is recommended, with an exponential decay curve for progressive load reduction. Taper strategies should be individualized, with daily CMJ monitoring for real-time load adjustment during the taper period.