R1617 FADE
Build the 16:00–16:59 ET RTH-close range. At 18:00, note the starting side of mid. First Asia-hours touch of mid from that side → fade back 15 points.
updated 2026-04-21The setup
The final 60 minutes of RTH (16:00–16:59 ET) builds a range. After the close, that range’s midpoint acts as a magnet overnight — and when price revisits the midpoint from the same side it started on, it fades.
- Start ABOVE mid at 18:00 → first touch of mid during Asia → LONG at mid, target mid + 15pt
- Start BELOW mid at 18:00 → first touch of mid during Asia → SHORT at mid, target mid − 15pt
Entry rules
- Range window: 16:00–16:59 ET (RTH close hour)
- Starting side: taken from the 18:00 ET bar
- Active window: 18:00 ET through 03:00 ET next day (Asia session)
- Trigger: first touch of the 16:17 range midpoint from the starting side
- Target: mid ± 15 points
- Stop: 30 points
- Time stop: 60 minutes after entry
- Range filter: skip if 16:17 range > 100pt (too volatile) or < 5pt (too tight)
Backtest
backtest_r1617_mid_fade.py — best config at tgt 15 / stop 30 / time_stop 60min:
| Metric | Value |
|---|---|
| Sample | n = 744 |
| Win rate | 67.1% |
| Avg R | +0.28 |
| Total points | +2,734 |
| Profit factor | 1.81 |
| Profitable years (2021–2026) | 6 / 6 |
Per-day-of-week: Monday best (PF 2.59), Wednesday weakest. No DOW is negative.
Why 60-min time stop
We tested 90-min and 120-min time stops — both worse (PF 1.65 / 1.66). The 60-min cap exits would-be-stops near breakeven instead of letting them walk all the way to the −30pt adverse level. A surprisingly clean example of how a tight time stop can beat a wider one.
The prior touch-study
Before running the fully simulated backtest, a simpler touch study showed 87% bounce rate on first mid touch from starting side (n=890). That number dropped to 67% once we required the bounce to complete to the +15 target with a 30-stop and 60-min time cap — but the pattern itself is real.
Why Tier 2
This is an Asia-session signal — it fires when RTH traders are asleep, lives and dies in the Asia window, and rarely bleeds into NY. It feeds the board during overnight reads but isn’t a day-trade primary.
Implementation
Session-date logic handles the wrap-around: if the current-bar hour is ≥ 17, the range is today’s; if it’s < 3, the range is from yesterday’s RTH close. Suppressed after the first touch fires — no re-entries.
History
- 2026-04-21 — Shipped. Backtest found no improvement in 90 / 120-min stops; 60-min optimal because it rescues would-be-stops.