📈 HMM Identification Regime-Based Portfolio Optimization Project

A comprehensive implementation of Hidden Markov Model (HMM) regime identification and regime-aware portfolio optimization strategies for enhanced risk-adjusted returns.

🎯 Project Overview

This project implements a sophisticated approach to portfolio management by:

1. Identifying economic regimes using Hidden Markov Models with leading economic indicators
2. Optimizing portfolio allocations dynamically based on the current economic regime
3. Comparing performance against traditional static portfolio optimization approaches

📊 Key Features

• HMM Regime Identification: 4-state economic regime classification (Expansion, Recovery, Downturn, Slowdown)
• Dynamic Portfolio Optimization: Regime-specific Mean Variance Optimization (MVO)
• Comprehensive Validation: Statistical and economic validation of regime models
• Performance Analysis: Detailed comparison of regime-based vs. traditional approaches
• Real Economic Events: Validation against known recession periods (1990s, Dot-com, Great Recession, COVID-19)

🏗️ Project Structure

Regime_Portfolio_project/
├── Data/
│   ├── leading_indicators.xlsx      # VIX, PMI, Yield Curve data (1990-2025)
│   ├── portfolio_data.xlsx          # Asset returns (Equities, Bonds, Commodities, REITs)
├── Regime_identification/
│   ├── hmm_regime_identification.ipynb    # Core HMM implementation
│   ├── hmm_validation.ipynb              # Model validation & testing
│   └── hmm_regime_identification.xlsx    # Generated regime classifications
├── full_period_mvo/
│   ├── mvo_normal.ipynb             # Traditional static MVO baseline
│   ├── hmm_regime_mvo.ipynb         # Regime-based dynamic MVO
├── Traditional_mvo_train_test/
│   ├── mvo_static.ipynb             # Static optimization with train/test split
│   ├── hmm_regime_mvo.ipynb         # Regime-based with train/test split
│   └── probabilities_regime.ipynb   # Regime probability analysis
├── HMM_Regime_Report_Comprehensive.ipynb  # Complete analysis report
└── HMM_Validation_Report_Comprehensive.ipynb  # Validation summary

🔬 Methodology

1. Economic Regime Identification

Input Features:

• VIX: Market volatility index (fear gauge)
• PMI: Purchasing Managers' Index (economic activity)
• Yield Curve: 10Y-2Y Treasury spread (economic outlook)

HMM Configuration:

• 4 hidden states (optimal by BIC criteria)
• Gaussian emissions with full covariance
• Standardized features for numerical stability

Regime Classification Logic:

• Expansion: Highest PMI, lowest VIX (strong growth, low volatility)
• Recovery: High yield curve, moderate VIX (economic recovery phase)
• Downturn: Lowest PMI, highest VIX (recession/crisis periods)
• Slowdown: Moderate indicators (economic deceleration)

2. Portfolio Assets

Asset Classes:

• EQU: S&P 500 (Equities)
• GOVT: Treasury Bonds (Government bonds)
• CORP: Corporate Bonds
• COMM: Commodity Index
• REITs: Real Estate Investment Trusts

3. Optimization Framework

Mean Variance Optimization (MVO):

• Utility function: U = μ'w - (λ/2)w'Σw
• Risk aversion parameter: λ ∈ [1, 10]
• Dynamic bounds based on inverse volatility
• Regime-specific expected returns and covariance matrices

📈 Key Results

Model Validation

• ✅ Optimal Model Selection: 4-state model chosen by BIC criteria
• ✅ High Regime Persistence: 93.5% average persistence rate
• ✅ Economic Validity: 100% recession detection success rate
• ✅ Realistic Duration: Average regime length of 19.4 months

Economic Event Validation

Event	Period	Detected Regime	Accuracy
Early 1990s Recession	1990-1991	Downturn (87.5%)	✅
Dot-com Recession	2001	Downturn (100%)	✅
Great Recession	2007-2009	Downturn (100%)	✅
COVID Recession	2020	Slowdown (100%)	✅

Regime Distribution (1990-2025)

• Recovery: 34.3% (12.1 years)
• Expansion: 31.9% (11.3 years)
• Downturn: 22.9% (8.1 years)
• Slowdown: 10.9% (3.8 years)

🚀 Getting Started

Prerequisites

pandas>=1.3.0
numpy>=1.21.0
scikit-learn>=1.0.0
hmmlearn>=0.2.7
scipy>=1.7.0
matplotlib>=3.4.0
seaborn>=0.11.0

Installation

git clone https://github.com/JATINDHURVE/Regime-Identification-using-Hidden-Markov-Model-and-Portfolio-optimization.git
cd regime-portfolio-optimization
pip install -r requirements.txt

Usage

1. Run Regime Identification:

# Execute the HMM regime identification
jupyter notebook Regime_identification/hmm_regime_identification.ipynb

2. Validate Model:

# Run comprehensive validation
jupyter notebook Regime_identification/hmm_validation.ipynb

3. Portfolio Optimization:

# Compare regime-based vs traditional approaches
jupyter notebook full_period_mvo/hmm_regime_mvo.ipynb
jupyter notebook full_period_mvo/mvo_normal.ipynb

📋 Notebooks Description

Core Analysis

• hmm_regime_identification.ipynb: Main HMM implementation with regime classification
• hmm_validation.ipynb: Comprehensive model validation and statistical tests
• mvo_normal.ipynb: Baseline traditional portfolio optimization
• hmm_regime_mvo.ipynb: Dynamic regime-based portfolio optimization

Validation & Testing

• mvo_static.ipynb: Static optimization with train/test methodology
• probabilities_regime.ipynb: Regime probability analysis and transitions

Reports

• HMM_Regime_Report_Comprehensive.ipynb: Complete project analysis
• HMM_Validation_Report_Comprehensive.ipynb: Detailed validation results

🔍 Key Findings

Regime Characteristics

1. Expansion Regime: Low volatility (VIX), high economic activity (PMI)
2. Recovery Regime: Steep yield curve, moderate volatility
3. Downturn Regime: High volatility, low economic activity
4. Slowdown Regime: Moderate indicators, economic uncertainty

Portfolio Implications

• Dynamic allocation based on economic conditions
• Risk management through regime-aware positioning
• Enhanced returns via tactical asset allocation
• Reduced drawdowns during crisis periods

📊 Performance Metrics

The regime-based approach demonstrates:

• Improved risk-adjusted returns
• Better downside protection during recessions
• More responsive allocation to economic conditions
• Enhanced portfolio diversification benefits

🤝 Contributing

Contributions are welcome! Please feel free to submit a Pull Request. For major changes, please open an issue first to discuss what you would like to change.

📄 License

This project is licensed under the MIT License - see the LICENSE file for details.

📚 References

• Bloomberg. (2023). "Bloomberg Fixed Income Indices: US Corporate Index"
• Bloomberg. (2024). "Bloomberg Fixed Income Indices: Bloomberg US Treasury Index"
• Bloomberg. (2025). "Bloomberg Commodity Indices: Bloomberg Commodity Index (BCOM)"
• Bouvé, E. & Teiletche, J. (2024). "Regime-Based Strategic Asset Allocation"
• Chan, E., Fan, H., Sawal, S. & Vialle, Q. (2023). "Conditional Portfolio Optimization: Using Machine Learning to Adapt Capital Allocations to Market Regimes"
• Chen, Y.M., Li, B. & Saunders, D. (2025). "Exploratory Mean-Variance Portfolio Optimization with Regime-Switching Market Dynamics"
• Costa, G. & Kwon, R.H. (2019). "A Regime-Switching Factor Model for Mean–Variance Optimization"
• Gómez-Cram, R. (2021). "Late to Recessions: Stocks and the Business Cycle"
• Kim, M.J. & Kwon, D. (2022). "Dynamic Asset Allocation Strategy: An Economic Regime Approach"
• Moench, E. & Stein, T. "Equity Premium Predictability Over the Business Cycle"
• OECD. (2020). "Interpreting OECD Composite Leading Indicators (CLIs)"
• Oliveira, D.C., Sandfelder, D., Fujita, A., Dong, X. & Cucuringu, M. (2025). "Tactical Asset Allocation with Macroeconomic Regime Detection"
• Pomorski, P. & Gorse, D. "Multi-period Portfolio Optimisation Using a Regime-Switching Predictive Framework"
• Shu, Y. & Mulvey, J.M. (2024). "Dynamic Factor Allocation Leveraging Regime-Switching Signals"
• Shu, Y., Yu, C. & Mulvey, J.M. (2024). "Dynamic Asset Allocation with Asset-Specific Regime Forecasts"
• Wang, M., Liu, Y.-H. & Mikkelson, I. (2020). "Regime-Switching Factor Investing with Hidden Markov Models"
• Yan, X. "Optimal Portfolio Allocation with Regime-Switching and Jump-Diffusion Dynamics: A Deep Reinforcement Learning Approach"

---

Note: This project is for educational and research purposes. Past performance does not guarantee future results. Always consult with financial professionals before making investment decisions.