✨ Feature Overview
FxMath AI Evolution combines cutting-edge machine learning with professional-grade trading tools to deliver a fully autonomous trading experience.
🧠 AI Engine
- LightGBM Core: State-of-the-art gradient boosting for superior predictive accuracy.
- Online Learning: Continuous model adaptation from live market data — retrains every 10 candles.
- SL/TP Hit-Test Labeling: Models are trained to predict whether TP (3× ATR) or SL (1.5× ATR) will be hit first, aligning ML training with actual trade outcomes.
- Confidence Scoring: Smart risk/reward assessment with configurable confidence thresholds.
- 12 Technical Features: RSI, MACD, ATR, ADX, Bollinger Bands, SMAs, and EMAs for comprehensive market analysis.
📊 Trading
- Multi-Timeframe: Simultaneous M5, M15, M30 analysis with voting-based signal aggregation.
- Multi-Symbol: Trade unlimited pairs simultaneously through independent instances.
- Signal Voting: Requires 2 out of 3 timeframe confirmations to filter out noise.
- Instant Execution: Lightning-fast order placement via MT5.
- Prediction Horizon: 10-bar forward SL/TP hit-test simulation for robust label generation.
🛡️ Risk Management
- ATR-Based Stops: Dynamic SL = 1.5× ATR, TP = 3.0× ATR for adaptive volatility protection.
- Trailing Stop: Automatic profit locking as the position moves in your favor.
- Breakeven: Zero-loss trade protection once a configurable profit threshold is reached.
- Position Sizing: Fixed lot or percentage-based risk per trade (default: 1%).
- Smart Filters: Time-based trading windows and news event blackout periods.
⚙️ Configuration
- Simple Setup: Standalone portable executable — no Python or dependencies needed.
- Auto-Detection: Seamless MT5 terminal discovery and integration.
- Strategy Optimizer: Built-in walk-forward optimization with random search across 20+ parameters.
- SQLite Database: Persists optimized settings per symbol/timeframe for quick reload.
- Real-Time Dashboard: Live monitoring of instances, signals, and trading performance.
Technical Specifications
| Metric | Value |
|---|---|
| Prediction Speed | < 10ms |
| Execution Speed | < 100ms |
| Memory Usage | 200–400 MB per instance |
| Min RAM | 4 GB (8 GB recommended) |
| Supported Timeframes | M5, M15, M30 (configurable) |
| ML Model | LightGBM (binary classifier) |
| Prediction Horizon | 10 bars |
| Retrain Interval | Every 10 candles (configurable) |