Back to Projects
Random ForestGaussian CopulaMHCICTAlpha GenerationMulti-AssetPythonQuantitative FinancePortfolio Optimization

Financial Forecasting System

A comprehensive quantitative trading system implementing Random Forests and Gaussian Copulas for alpha generation across multiple asset classes.

Gaussian Copula
Statistical Models
Mixture of Experts
Style
31 (Equities, Crypto, Forex)
Assets

# Methodology

• Random Forest Model: Implements 100-tree ensemble learning with 15 engineered features including moving averages, RSI, MACD, Bollinger Bands, ATR, and volume indicators. The model achieves 62% 1-month hit rate with risk-adjusted alpha generation across 31 assets.

• Gaussian Copula: Models asset dependencies and correlations using 10,000 Monte Carlo simulations. Captures non-linear relationships and tail dependencies between assets, providing robust portfolio optimization and risk management insights.

• Manifold Constrained Hierarchical Clustering: Uses Isomap embedding to preserve geodesic distances in 3-5 dimensional manifold space. Applies sector coherence constraints and momentum persistence analysis for improved clustering stability.

• ICT Integration: Incorporates Institutional Trading Concepts including liquidity zones, order flow analysis, order blocks, and kill zones. Provides precise entry/exit levels with risk-reward ratios of 1:1.3 to 1:2.5.

Core Model

Random Forest + Gaussian Copula + MHC + ICT

# Data Source

Yahoo Finance, Alpha Vantage, Real-time Market Data