Two antagonistic models are used to model the training-performance relationship during an eight week cycling training, with a prediction of the performances in the last week of training. Results indicated different qualities of the model fits for the test data for both models. Mean differences between predicted and real performances were 2.39% for the Fitness-Fatigue-Model and 4.43% for the PerPot-Model. However, the Fitness-Fatigue-Model parameters remain questionable in terms of physiological interpretation. Hence, the PerPot-Model is preferred, although the general applicability for performance prediction in the presented cycling training situation can not be supported.
© Copyright 2006 International Journal of Computer Science in Sport. Sciendo. All rights reserved.