2: I might also play around with crash and correction frequencies if you’re not already doing so, which, without thinking deeper about credit cycles, should be fine to simulate based on previously occurring frequencies.

It seems like you already have some sort of analog to this with historical performance to begin with, though.