The point being, historical statistics just don't tell you everything you need to make a decision…you need guarantees that all regimes have outcomes within acceptable parameters.

Accounting for Murphy's law, and that the worst case is always x% below your expectation (even after accounting for Murphy's law).