Momentum Gate NAS100
- Experts
-
Aaron Touboul
XAU/USD Trader (SMC/ICT) for ~2 years and MQL5/MT5 Developer
I trade gold on a daily basis, so I understand what an SMC logic must do at execution level — not just how to code it.
Specialties:
* XAU/USD EAs based on SMC/ICT methodology: market structure, FVG, OB, liquidity - Version: 5.1
- Activations: 10
Four features were removed from this EA because they lost money.
Six trades a month. Three filters reject everything else.
Most NASDAQ expert advisors are sold on a backtest curve. This one is sold on what was taken out of it.
THE TRIGGER IS WORTHLESS ON ITS OWN
Two consecutive M15 candles in the same direction, each with a body of at least 26 points — half an hour in which price advances without hesitating. Measured alone, that pattern returns +0.061 R in discovery and −0.021 R in validation. It loses money. Everything this product is worth sits in what comes next.
THREE REFUSALS
Each filter was added because it was measured, not because it sounded sensible.
- Is the move genuinely abnormal? A 26-point body is ordinary on a busy day and significant on a quiet one. The EA enters only if current volatility is below 1.50x its background volatility. Without this filter: 924 trades at +0.163 R. With it: 537 trades at +0.294 R.
- Is the move already over? An impulse larger than 2.5 ATR in thirty minutes means price has already travelled two and a half times its normal range. That is buying the top. Rejected.
- Are we too far from the base? If price has already moved more than 3.3 ATR from its 20-bar base, the trade is rejected.
A fourth condition filters direction: price must already be moving the way of the trade over the last 20 M15 bars, about five hours. The EA never takes counter-trend.
Everything is evaluated on the M15 close. No decision is taken inside a bar, so nothing repaints.
FOUR WAYS OUT, EACH MEASURED SEPARATELY
Most EAs have a stop and a target. This one has four doors.
- Stop: below the extreme of the last 8 M5 bars, so on real structure rather than an arbitrary distance. Measured plateau 7 to 10 bars; the result collapses at 4 and at 16.
- Target: 2x risk. Plateau 2.0 to 2.25. For reference, 1:1.5 gives +0.088 R and 1:4 gives +0.056 R — both still positive, which means the setting does not hang on a single value.
- Partial exit: at +1.5 R half the position closes and the stop moves to entry. From that point the trade cannot lose.
- Time stop: if after 12 M15 bars (three hours) the trade has never reached +0.5 R, it is closed. Plateau 8 to 20 bars.
A fifth rule, optional, closes a trade that pulls back more than 0.75 R without ever working: trades that retrace that far do not come back (18 trades beyond 0.80 R averaged −0.96 R). It comes from chart observation rather than a grid, rests on 83 trades, and its plateau is narrow. It is a documented option you can switch off, not a pillar.
HARD LIMIT: TWO TRADES A DAY
Days with four or five trades measured −2 to −3 R. This is not a comfort rail, it is a filter that makes money.
WHAT THIS EA DOES NOT DO
No martingale. No grid. No averaging down. No pyramiding — pyramiding was coded, measured at −0.125 to −0.154 R per trade (t = −23 to −43), and removed.
WHAT WAS REMOVED, AND WHY
- Pyramiding: −0.125 to −0.154 R per trade.
- H4 bias: measured internally at +0.19 R, then refuted by a real MT5 backtest. The effect came from approximating H4 with 16 M15 bars, and disappeared with true H4 bars.
- Session-close exit: it broke the fixed ratio and closed half the trades flat.
- Waiting for a pullback before entry: degraded both years. On momentum, a pullback only returns on the moves that failed.
WHY THE SETTINGS ARE WHAT THEY ARE
Every published setting is the centre of a zone where the result stays positive, not the best point of a grid. Stop: positive from 7 to 10 M5 bars. Target: positive from 1.5 to 2.5 R. Volatility gate: positive from 1.05 to 1.80. Grid sweeps: 70 positive cells out of 80 on extension, 20 out of 20 on impulse amplitude. An overfitted EA sits on a narrow peak; this one sits on plateaus.
BACKTEST 2025–2026 — USTEC M15, 1% risk per trade
| Measure | Value |
|---|---|
| Sample | 145 trades over 23 months |
| Expectancy | +0.346 R per trade |
| 2025 tested alone | +0.278 R |
| 2026 tested alone | +0.375 R |
| Win rate | 53% |
| Positive months | 14 of 19 |
| Maximum drawdown | 3.8 R |
| MT5 backtest, 3 Nov 2026 | profit factor 1.46, recovery factor 2.67, drawdown 4.09% |
Both years were tested separately and both are positive. These are backtest figures, not a forecast of future returns.
WHAT HAS NOT BEEN MEASURED
Stated here because a serious buyer would find it anyway:
- One instrument, two years. Nothing says how the EA behaves on another index, or before 2025.
- 145 trades is a small sample. At a 53% win rate, the confidence interval on the true rate spans roughly 45% to 61%.
- There is no true out-of-sample period. 2025 and 2026 were tested separately, which is better than nothing, but both were visible while the parameters were chosen.
- No live track record yet.
One precedent that should make you cautious, and that is why it is printed here: the H4 bias above was measured internally at +0.19 R on a plateau, across both years, and a real MT5 backtest then refuted it. That happened once on this product. It can have happened elsewhere without being caught.
TEST IT YOURSELF FIRST
The free demo runs in the Strategy Tester. Download it, run the years and the broker data you care about, and judge the product on your own results rather than on mine. Questions through the product comments.
