Many energy system sizing tools are based on Mathematical Programming (MP) formulations that implicitly assume perfect foresight for the underlying system operation (also called energy management or unit commitment and energy dispatch). In contrast, sizing tools that rely on a system simulator embedding simple rule-based operation strategies are admittedly suboptimal, but are non-anticipative by default. Using the open-source simulator-based sizing tool Microgrids.jl and a MP-based variant, we study the risk of obtaining an ill-sized microgrid (i.e. one that cannot meet a prescribed Quality of Service level) when assuming perfect foresight.