🧠 GOLD EA (XAUUSD M15 · ONNX MODEL EMBEDDED)
GOLD NEURON EA
Trade direction is decided by an AI model embedded in the EA (ONNX gradient boosting, 13 market features). One decision per day at a fixed London-session time (04:15 UTC, DST-aware); every position carries a fixed TP/SL bracket from the moment it opens. 3 split positions, no martingale, no grid, no averaging. No external files, no DLLs — works offline.
🧠 About the AI model's training period — read before use
The embedded model is trained on data through July 2026. Backtests inside the trained period (all figures below) measure the model's fit to that history — they are not a forward guarantee. The model is retrained periodically and shipped as free version updates; always run the latest version.
⚠ Aggressive EA (high risk) — read before use
The default mode (Ultra) targets maximum growth and reached a measured ~79% equity drawdown. Recommended: Aggressive (growth) or Defensive (safety). Trade only with money you can afford to lose, and always start on demo or a small account.
⚡ Minimum deposit: from $100
XAUUSD, any chart timeframe (signals are computed on M15), leverage 1:500 recommended. Compounding percent-of-equity sizing follows your balance automatically. For small/safe accounts use a micro (cent) account with Defensive mode.
Profit Factor
2.83
期待値
勝率
89.3%
447取引
最大DD
42.40%
証拠金DD
年率複利
+79632.0%
+$7,963,200/年（$10,000基準）
Attach to: XAUUSD M15 — The strategy reads the M15 timeframe. On any other chart it will place no trades at all. When testing in the MT5 Strategy Tester, set the symbol and timeframe to XAUUSD M15 as well.
HOW IT WORKS — EMBEDDED AI DIRECTION MODEL
How it works — embedded AI direction model
A gradient-boosting classifier evaluates 13 market features (multi-horizon momentum ladder, intraday range, calendar context) and decides the day's direction — once per day. After entry, everything is left to the fixed bracket.
- Entry: once per day at 04:15 UTC (London session, DST-aware). Market entry in 3 splits following the AI direction
- Exit: every position carries a fixed TP/SL bracket from the moment it opens. No discretion, no adds, no averaging — bracket only
- Risk: compounding percent-of-equity sizing with margin and volume-limit guards. A single RunMode dropdown scales risk across 4 presets
- The model is embedded in the .ex5 (ONNX). No external files, no DLLs, no API calls — works offline. Periodic retrains ship as free updates
ONE DROPDOWN — FOUR RISK PRESETS
RunMode dropdown — four risk presets
| Preset | Risk × | Return (real tick, 2025 full year) | Max equity DD |
|---|---|---|---|
| Defensive | 0.5 | +688% | 14.6% |
| Standard | 1.0 | +5,170% | 27.7% |
| Aggressive | 2.0 | +75,302% | 22.9% |
| Ultra (default) | 2.5 | +102,363% | 52.4% |
All presets share the same signals — only risk scales. Real tick, XAUUSD M15, $10,000 start, 1:500, default inputs. 2026 (Jan–mid Jul) is consistent: Defensive +341%, Standard +1,752%, Aggressive +19,369%, Ultra +—% / 76.5% DD. Standard-mode yearly 2020–2024: +261% to +2,465% (every tested year positive). All figures are inside the model's trained period (see disclaimer above).
VALIDATION — OTHER SYMBOLS & OTHER BROKERS
Validation — other symbols & other brokers
Research models built with the same 13-feature architecture, trained per symbol, were positive in all 21 yearly real-tick windows 2020-2026 on EURUSD, GBPUSD and USDJPY (MT5, $10,000, 1:500). This product trades XAUUSD only; these are separate research models.
| Symbol | Positive windows | Best year (real tick) |
|---|---|---|
| EURUSD | 7/7 | 2022: +12,534% / PF 3.56 |
| GBPUSD | 7/7 | 2022: +16,006% / PF 4.66 |
| USDJPY | 7/7 | 2022: +14,327% / PF 2.26 |
Cross-broker check (GMT+0, wider-spread standard-account feed, 2017-2026): Standard mode was positive in every tested year inside the training window (2020-2026: +221% to +3,450% per year). 2017 and 2019 - outside the training window - were losing years, confirming the trained-period caveat. Ultra reached 92-99% equity drawdown on that feed - use Standard or Defensive there. On fixed-offset servers set GMTWinterOffset=0 and AutoDST=false (v1.32+).
* All figures are backtests inside each model's training window and do not guarantee future results.
📈MT5 Strategy Tester 報告 — GOLD NEURON EA
📋 關於此報告
- MT5 策略測試器(XMTrading-MT5 伺服器)生成的官方報告
- 5-10 年實際價格資料 + 全 tick(99.9% 精度)模型
- 餘額曲線(藍)/ 淨值曲線(綠)可視化浮動損失
- 過去結果不保證未來利潤
⚠ Where it fails — honest limits
- All published figures are backtests inside the model's training window (through July 2026). Out-of-training performance is not guaranteed and a static model decays over time — applying the periodic retrain updates is assumed
- Default Ultra reached a measured ~79% equity drawdown — nearly 80% of the account underwater. If that is unacceptable, choose Aggressive, Standard or Defensive
- Ultra (×4) is the measured risk ceiling: ×5 and above blew up to 85–95% eqDD in testing. Raising the risk inputs beyond defaults risks account loss
- Backtests are past results and do not guarantee future profits. Live track record is accumulating
下載（免費）
購買 GOLD NEURON EA
Buy on the official MetaQuotes Market (MQL5 Market). Payment, licensing and automatic updates are all handled inside the MT5 app.
🏷 現為首發價 —— 每售出10份價格上調一檔。越早購買越優惠。
Buy on MQL5 Market →Buy on the official MetaQuotes Market — install directly in MT5 with automatic updates. No account registration needed.
* Figures are real-tick backtest measurements (MT5 Strategy Tester, default inputs, inside the model's trained period). They do not guarantee future profits. Automated FX trading can lose your principal. Use risk capital only.