Hercules is a dwarf spheroidal satellite of the Milky Way, found at a distance of about 138 kpc, and showing evidence of tidal disruption. It is very elongated and exhibits a velocity gradient of 16 +/- 3 km/s/kpc. Using this data a possible orbit of Hercules has previously been deduced in the literature. In this study we make use of a novel approach to find a best fit model that follows the published orbit. Instead of using trial and error, we use a systematic approach in order to find a model that fits multiple observables simultaneously. As such, we investigate a much wider parameter range of initial conditions and ensure we have found the best match possible. Using a dark matter free progenitor that undergoes tidal disruption, our best-fit model can simultaneously match the observed luminosity, central surface brightness, effective radius, velocity dispersion, and velocity gradient of Hercules. However, we find it is impossible to reproduce the observed elongation and the position angle of Hercules at the same time in our models. This failure persists even when we vary the duration of the simulation significantly, and consider a more cuspy density distribution for the progenitor. We discuss how this suggests that the published orbit of Hercules is very likely to be incorrect. ; Comment: accepted by MNRAS; 19 pages, 19 figures, 2 tables