Run repeated stochastic trials at a chosen population size and watch allele-frequency trajectories diverge (or not).