Goldfinger AI combines multi-horizon H1 liquidity clustering, M5 microstructure hysteresis, and an embedded 35-feature Dual-MLP gate trained directly on MQL5 price-action exports (2019–2026). Zero Martingale. Zero Grid. Zero DLLs.
Every curve, yearly breakdown, and vector metric below is parsed directly from the Strategy Tester HTML reports in this repository. Toggle account deposit ($500 / $1,000 / $10,000) and neural profile (NN1 / NN2).
| Year | Trades | Win Rate | Net Profit (incl. Comm & Swap) | End Balance |
|---|
No naked EMA or RSI crossovers. Every candidate entry passes through Engine I (Spatial Liquidity Topology), Engine II (M5 Hysteresis), and a 35→64→32→1 neural probability filter.
Test how the embedded neural probability gate (P >= 0.52) and Engine I/II structural guards approve real momentum setups and veto liquidity traps.
Built-in TesterHideIndicators(true) keeps MT5 charts clutter-free while rendering real-time neural confidence, liquidity pool distances, and risk budget directly on canvas.
Calibrated for ECN / Raw-spread accounts from $500 micro accounts (0.01 min-lot risk fallback) up to $10,000+ institutional compounding.
| Specification | Goldfinger AI (XAUUSD) | Bitfinger AI (BTCUSD) |
|---|---|---|
| Chart Timeframe | H1 (reads M5 internally) | H1 (reads M5 internally) |
| Default Magic / Prefix | 20260 / GFAI_ | 20268 / BFAI_ |
| Account Type | ECN / Raw Spread (Hedging or Netting) | ECN / Raw Spread Crypto CFD (Hedging or Netting) |
| Daily DD Circuit Breaker | 3.5% Intraday Equity Limit | 4.5% Intraday Equity Limit + Weekend Block |
| Minimum Lot / Fallback | 0.01 lot ($500+ Balance) | 0.01 lot ($500+ Balance) |
Both Goldfinger AI and Bitfinger AI launch with identical transparent MQL5 Market pricing, including NN1 + NN2 profiles, lifetime updates, and HUD.
Download the Free MT5 Demo on MQL5 Market to run Strategy Tester on your broker feed, rent via MQL5, or save with Direct Site Licensing below.
Never. Every single position opens with an independent hard Stop Loss, predefined Take Profit or dynamic velocity trailing, single-direction net exposure locking, and an intraday equity drawdown circuit breaker.
Zero external DLLs are required. Both Dual-MLP neural networks (`35→64→32→1`) and z-score scaling constants are compiled natively inside self-contained MQL5 `.mqh` headers, passing MQL5 Market automated validator checks with zero dependencies.
NN1 (`InpExecArchitecture = 0`) targets higher trade frequency with tighter protection and dynamic trailing / 0.60R velocity captures. NN2 (`InpExecArchitecture = 1`) selects wider structural swing stops and 1.50R expansion targets with lower trade count and higher profit factor.
Strategy Tester reports are simulated historical backtests (`99%` history quality) showing how the architecture behaved over 2019–2026 (Gold) and 2022–2026 (BTC). Live execution depends on broker spread, slippage, liquidity, and VPS latency — monitor live Magic 20260 / 20268 execution cards in our Telegram channel.
Yes. Both EAs include verified `$500` and `$1,000` Strategy Tester reports using `0.01` minimum lot fallback (`InpAllowMinLot = true`) and `MinLotMaxRiskPct` guards. For running both EAs concurrently with dynamic equity scaling, `$1,000+` per symbol is recommended.