feat: Add machine learning arbitrage prediction models

- Implement Random Forest, Gradient Boosting, XGBoost, and LightGBM models
- Advanced feature engineering with market data and historical performance
- Neural network predictor using PyTorch for complex patterns
- Ensemble prediction combining multiple ML models
- Model training, evaluation, and performance tracking
- Feature importance analysis and model persistence
- Real-time prediction API for arbitrage opportunities

Author: Andrew Leng

requirements.txt

@@ -8,3 +8,9 @@ numpy>=1.21.0
 pandas>=1.3.0
 python-socketio>=5.0.0
 eventlet>=0.33.0
+scikit-learn>=1.0.0
+xgboost>=1.6.0
+lightgbm>=3.3.0
+torch>=1.12.0
+torchvision>=0.13.0
+joblib>=1.1.0