A clean quantitative framing is this: in a crisis, correlations converge — which means diversification fails exactly when you need it most. Desk note: Assets that appear uncorrelated in normal markets can move in lockstep during stress events. This makes naive diversification less protective than it appears. Why investors care: That is why stress-testing should use conditional correlations, not historical averages. Translate it into behavior: In March 2020, equities, credit and even gold initially sold off together as liquidity evaporated. The diversification that "worked" in 2019 failed in the first weeks of the crisis. Where people usually get tripped up: The mistake is building portfolio protection around average-condition math when tails are where the real damage happens. Keep this nearby on the next review: Write down the state variable you would monitor first if this thesis started to drift. The point is not to memorize the label. The point is to know what variable is actually doing the work.