This study aimed to evaluate the effect of simulated training strategies on performance potential in elite short-track speed skaters. Training load and field-based criterion performances from fifteen athletes (10 males, 5 females) were collected over a 3-month training period and the relationship between training loads and performance was computed with a variable dose-response model using a genetic algorithm. Individual simulations of tapers preceded or not preceded by an overload training (OT) were assessed. We obtained a significant correlation between actual and modelled performances (R ²  = 0.76 ± 0.07). Regarding model parameters, no significant difference was found between males and females but the time to recover performance tended to be lower in females. Simulations in which the taper parameters were free highlighted that an exponential or a step taper were the most effective for increasing performance compared to a linear taper (p < 0.05). Optimal exponential taper duration after OT was 10.7 ± 2.4d and the optimal load reduction was 75.9 ± 3.7%. OT intensity had the greatest influence on the predicted performance, followed by OT duration, taper decay, and to a lesser extent load reduction during taper and taper duration. Thus, a variable dose-response systems model allows the evaluation of different taper strategies and their potential effect on performance changes.