I prefer Conditional Value at Risk/Expected Shortfall. If you’re working with simulations, it’s actually pretty easy to work with (though it is more of a pain to work with when the distributions are parametric). Part of the problem is that as an investor, you never really care about historical returns and risk, you care about expected returns/risk. Forward-looking and not backward-looking. I agree with Black Swan that all these things need to dynamically change as regimes change. Unfortunately, it can be quite difficult to actually do that, even when you can convince your boss of the value.