Sunmool's NY Lunch Model Backtesting

ICT NY Lunch Model Backtesting (12:00–13:00 NY) 🗽🍔

This research indicator tests an ICT narrative using the New York lunch window (12:00–13:00 America/New_York). It records that hour’s high/low and measures, during the post-lunch session (default 13:00–16:00), how often:

⬆️ If the afternoon trends up, the Lunch Low gets swept first.

⬇️ If the afternoon trends down, the Lunch High gets swept first.
It reports these as conditional probabilities, not trade signals. 📈

👀 What it shows

🟦 Lunch Range box (toggle): high/low from 12:00–13:00 NY

🔻🔺 Sweep signals (bar-anchored)

Low sweep: triangle below bar + optional “L”

High sweep: triangle above bar + optional “H”

🧱 Optional small box wrapping the swept candle

📊 Stats table (top-right)

P(L-swept | Up) — % of Up-days where Lunch Low was swept

P(H-swept | Down) — % of Down-days where Lunch High was swept

🔁 Contradictions + sample sizes (Up-days / Down-days)

🎯 Direction logic (Up/Down)

Anchor: 13:00 open (pmOpen) ⏰

Threshold: ATR × multiple or % from 13:00

Close ≥ pmOpen + threshold → Up-day

Close ≤ pmOpen − threshold → Down-day

Tiny moves under the threshold are ignored to reduce noise 🧹

⚙️ Inputs

🌐 Timezone: America/New_York (DST handled)

🍽️ Lunch window: 1200–1300

🕓 Post-lunch window: default 1300–1600 (try 17:00/20:00 for sensitivity)

📐 Trend threshold: ATR / Percent (with length/multiple or % level)

📅 Weekdays-only toggle (FX/Equities style)

👁️ Display toggles: Lunch box / sweep arrows / sweep text / sweep candle box / stats table

🔔 TF hint when chart TF > 15m

🧭 How to use

Use 5–15m charts for accurate lunch range capture.

Scroll ~1 year for meaningful samples.

Run sensitivity checks: vary ATR/% thresholds and the post-lunch end time.

For crypto, compare with vs without weekends. 🚀

🧠 Reading the results

High P(L-swept | Up) with a solid Up-day count ⇒ on up afternoons, lunch low is often swept.

High P(H-swept | Down) ⇒ on down afternoons, lunch high is often swept.

Lower Contradictions = cleaner tendency.

Remember: this is a probabilistic tendency, not a rule. 🎲

📝 Notes & limits

All markers (arrows, text, sweep boxes) are bar-anchored; the lunch range box is a research overlay you can toggle.

Real-time vs historical bar building can differ—interpret on bar close. 🔒