Faithful port of the user's submitted rule, settled in the idea builder (venue FX, 1h bars, long+short, leverage 1.5x, OHLCV only). Instrument: EURUSD.FX spot forex, 1-HOUR bars (~2 years of hourly history; Yahoo prices). Signal: 20-bar simple mean and 20-bar population std of hourly closes; upper band = mean + 2*std, lower band = mean - 2*std. Session filter: entries only on bars that open at or after 07:00 and close at or before 16:00 Europe/London local time (DST-aware, i.e. bars 07:00-08:00 through 15:00-16:00 London). A close above the upper band opens a short; a close below the lower band opens a long. Exits: the close returns to (crosses or touches) the current 20-bar mean, OR 12 hourly bars have elapsed since entry, OR an intrabar protective stop at 1.5 x ATR(20) from the entry fill price is hit. Execution substitutions (per the capability registry): FX is a session venue, so a decision on bar N's close fills at bar N+1's open (next_open_fill, supported); the stop is a platform-matched intrabar stop placed via submit_stop_order (reduce_only) — it triggers inside the bar and fills at the trigger or at a gapped open; it is not an exchange-resting order, which the idea does not ask for. Sizing ('risk-sized relative to account equity', completed with standard choices): notional = risk_pct x equity / stop_distance_pct, capped at leverage x equity with leverage = 1.5 (the user's settled value, below the 5x the text mentions as an upper bound). Cost arithmetic: hourly EURUSD std is ~0.08-0.12%, so a 20-bar 2-sigma excursion sits ~0.4-0.6% from the mean; if the trade recovers about a third of that distance on average after stops and time exits, gross capture is ~0.12-0.2% per trade versus a round trip of ~0.01-0.04% (0.2 bp commission per side + half the measured hourly spread per fill) plus ~0.015%/night overnight carry when a position is held through the 17:00 New York roll (late-session entries with a 12-bar clock can cross it). The platform's analyst fee floor for FX is 0.10% of notional per trade, so the honest expectation is near that floor — this is a real risk, stated up front, not a reason to change the user's rule. One prior FX experiment exists in the corpus; this fills the FX venue, long_short and mid-horizon buckets.