We Coded All Five Setups From a Viral ICT Strategy Guide. One Survived (Barely).

The Conclusion

Below: every trade of the surviving variant, then the five rule sets as we coded them, and the full results on both markets.

All 249 Trades of the Survivor, for Review

FVG retrace on 15-minute NQ: three-candle fair value gap whose middle candle body is at least 1.5× ATR(14), limit entry at the near gap edge on the retrace (within 48 bars), stop at the far gap edge, fixed 3R target, one position at a time. Filter the table, check any fill against your own charts — the signal timestamp is the bar that completed the gap.

The Source: Five Setups, Zero Numbers

The guide is a well-written tour of Inner Circle Trader vocabulary, demonstrated on BYBIT:BTCUSDT.P: fair value gaps and their “equilibrium,” the accumulation–manipulation–distribution arc of the Power of Three, filled gaps inverting into support and resistance, breaker blocks left behind by failed liquidity sweeps, and the three Silver Bullet hours (03:00–04:00, 10:00–11:00, 14:00–15:00 ET). Like most ICT material, it contains plenty of annotated hindsight charts and not a single win rate, expectancy, or trade count.

Narrative rules can't be tested, so we pinned down every blank — retrace windows, session anchors, stop placement, targets — with the most standard mechanical reading of each concept, wrote them as code, and applied the same conservative fill model we use in every backtest: limit entries, and when a bar touches both stop and target, the stop fills first.

Round 1: Bitcoin Perpetuals — the Guide's Own Market

Binance USDT-perp data: 1-hour bars from January 2022 (39,920 bars) for the swing setups, 5-minute bars from July 2024 (216,382 bars) for the session setups. Costs: 7.5 bps of notional per round trip — maker entry, taker exit at standard non-VIP fees. Best variant per strategy:

Notice the shape of the failure: pre-cost, three setups hover near zero — after costs, everything is red. The guide demonstrates these entries on the one market where their stop geometry cannot pay the toll. A 1-hour BTC fair value gap is often $20–50 wide; the round trip costs ~$45 at $60,000. No entry logic survives paying 0.5–2R per trade in fees.

Round 2: NASDAQ Futures — Where Costs Are Fair

Same five setups, 1-minute NQ data resampled to 5- and 15-minute bars: six months of Databento data (September 2025 – March 2026, 183,747 bars) plus a fresh TradingView export (June 29 – July 22, 2026). Costs: 0.5 points round trip against median stop distances of 15–57 points — negligible, so this is the honest test of the ideas themselves.

Three setups cluster at breakeven — the market-microstructure equivalent of a shrug. Inversion FVGs keep losing at scale. And the displacement-filtered FVG retrace repeats its positive result: +0.12R over 221 Databento trades, then +0.41R over 28 trades on the July weeks it had never seen.

The Survivor, Month by Month

Combined across both NQ samples: 249 trades, 29.7% winners against a 25% breakeven at 3R, +38.5R net. Here is the monthly equity, because averages hide the part you'd actually have to live through:

Six of eight periods positive — but December handed back −7R, and more than 70% of the total profit comes from January onward. We have seen this exact pattern before: our NY 8:30 session-strategy backtest also flipped from losing to winning in January 2026. Either NQ's intraday retrace behavior changed this year, or we are looking at a regime that will leave as quietly as it arrived.

And the multiple-testing tax applies: across this study we ran roughly 130 variant combinations. The best cell of 130 tries should look good by chance alone. One standard error of separation from zero does not clear that bar — which is why the rules are now frozen exactly as coded, awaiting a multi-year 1-minute NQ history for a true out-of-sample verdict.

Method Notes

• Conservative fills throughout: limit entries at the quoted level; when one bar touches both stop and target, the stop is assumed to fill first; entries gapping through the stop are counted as immediate losses.
• One position at a time per strategy; no pyramiding, no re-entries while a trade is open.
• Costs: 7.5 bps of notional per BTC round trip (maker entry, taker exit); 0.5 NQ points per futures round trip (commission plus one tick of slippage).
• Session logic is DST-aware New York time. The two NQ samples leave a gap (March 12 – June 28, 2026) where we simply have no 1-minute data.
• Where the guide names no target, we swept R-multiples (1–3R) and report the best — generosity that makes the four failures more damning, not less.

This is research, not trading advice. Backtested performance — especially a 249-trade sample with known regime clustering — is not a prediction of future results.