Ten years from now reactor neutrino experiments will attempt to determine which neutrino mass eigenstate is the most massive. In this paper we present the results of more than seven million detailed simulations of such experiments, studying the dependence of the probability of successfully determining the mass hierarchy upon the analysis method, the neutrino mass matrix parameters, reactor flux models, geoneutrinos and, in particular, combinations of baselines. We show that a recently reported spurious dependence of the data analysis upon the high energy tail of the reactor spectrum can be removed by using a weighted Fourier transform. We determine the optimal baselines and corresponding detector locations. For most values of the $CP$-violating, leptonic Dirac phase $\ensuremath{\delta}$, a degeneracy prevents $\mathrm{NO}\ensuremath{\nu}\mathrm{A}$ and T2K from determining either $\ensuremath{\delta}$ or the hierarchy. We determine the confidence with which a reactor experiment can determine the hierarchy, breaking the degeneracy.