Top panel: The distribution of the projected misalignment angle, Δθ, as a function of the number of particles. The orange points depict the prediction from the MC, assuming only sampling noise. The orange shaded region is the 1σ scatter from the MC. Over plotted with error bars are the values of Δθ from haloes matched between the m₈, m₉ and m₁₀ resolution runs of the simulation. Bottom panel: The scatter of Δθ as a function of the number of particles. This is the scatter of the top panel, with the r.m.s errors calculated using bootstrapping. The scatter on the misalignment angle from the simulations are larger than our pure sampling noise model, due to the additional scatter arising when resolved low-mass substructure is excised from the host halo.