Nice post — and this relates closely to why I find “Probabilistic Maximal Loss” in insurance to be such a bad idea, here stated very clearly. (I think that simulation is an improvement, but still allows for incoherent questions - but that’s a different point for a different time.)