Introduction: Modelling the individual load-performance relationship to deduce information about convenient application of load and recovery phases to achieve optimal adaptation processes and estimate fitness and fatigue levels is essential for training theory. The existing linear and non-linear systems models abstract real-world settings and underlying physiology and need to be compared regarding their predictive accuracy, ability to copy adaptations and included variables (Taha & Thomas, 2003). Methods: Training load and performance of five elite swimmers (17-27y, 1f/4m) was assessed for 25 weeks by recording swimming kilometers (volume) weighted by different stress levels (intensity) and performing Semi-Tethered-Tests (3x20m freestyle, increasing resistance) twice a week. Data was analyzed and modelled using the Fitness-Fatigue-Model (FF-Model), the Performance-Potential-Model (PerPot) and the Performance-Potential-Double-Model (PerPot DoMo) (Perl & Pfeiffer, 2011). Mean absolute percentage error (MAPE) and intraclass correlation coefficient (ICC) were computed to estimate the goodness of fit (model-fit) i.e. compare empirical and simulated performance. Results: Resulting MAPEs ±SD and mean ICCs were 3.07±1.81% and .65 for the FF-Model, 2.94±1.23% and .72 for PerPot and 2.58±1.16% and .81 for PerPot DoMo. Discussion: The results indicate a slightly better goodness of fit for the PerPot DoMo in comparison to the FF-Model and PerPot, but no significance because of the high variance and small sample size. Two separate inputs (volume/intensity) enable PerPot DoMo to model performance more precisely than with one input (FF-Model, PerPot), though the model fit of all three models is appropriate regarding the performance`s dynamics and variations.
© Copyright 2016 21st Annual Congress of the European College of Sport Science (ECSS), Vienna, 6. -9. July 2016. Veröffentlicht von University of Vienna. Alle Rechte vorbehalten.