Financial Portfolio Optimization Using Operations Research

Overview

This repository contains the implementation of an advanced financial portfolio optimization system integrating time-series forecasting, risk-reward evaluation, and mathematical optimization techniques. The goal is to dynamically allocate investments across Gold, Bitcoin, Ethereum, and Bonds, considering risk constraints and leverage.

Features

• Time-Series Forecasting: Employs linear regression models to predict future asset prices based on historical data.
• Feature Engineering: Utilizes lagged variables and moving averages to enhance prediction accuracy.
• Risk-Reward Analysis: Generates buy/sell signals based on predicted price movements and calculates potential profits and losses.
• Constraint Optimization: Implements constraint programming models in MiniZinc to determine optimal capital allocation while respecting leverage and risk tolerance limits.
• Weekly Capital Updates: Automatically adjusts the portfolio based on real market outcomes, tracking performance and accuracy over time.

Workflow

The complete workflow includes:

1. Data Preparation (prepare_data_for_this_week.py)

   • Extracts recent market data (100-day window).

2. Preprocessing and Feature Engineering (preprocessing.py)

   • Prepares and scales data, generating lagged features and moving averages.

3. Forecasting (linear_regression.mzn)

   • Applies linear regression via MiniZinc to forecast future prices.

4. Risk-Reward Analysis (compute_risk_reward.py)

   • Calculates potential buy/sell profits and generates trading signals.

5. Portfolio Optimization (main_model.mzn)

   • Solves optimization problems to allocate investments optimally.

6. Capital Update (update_capital.py)

   • Updates portfolio values weekly based on actual market performance.

7. Performance Tracking

   • Records capital growth, prediction accuracy, and trade effectiveness.

Performance Evaluation

• Typical weekly iteration time: under 30 seconds.
• Forecast and optimization models provide near-real-time decision-making capabilities.
• Tracks metrics including Return on Investment (ROI), accuracy rates, and capital growth.

Visualization

Currently, the repository does not include built-in visualization tools or scripts. However, suggested visualizations for implementation include:

• Forecasted vs. actual asset prices
• Capital growth trajectories
• Weekly allocation breakdown
• Prediction accuracy trends
• Risk-reward profiles

Limitations and Future Improvements

• Consideration of nonlinear market dynamics through advanced models (e.g., ARIMA, Random Forest, Neural Networks).
• Incorporation of adaptive risk management and dynamic leverage constraints.
• Inclusion of transaction costs and detailed margin rules.
• Real-time data streaming integration for intraday trading.

Installation and Usage

Clone the repository:

git clone https://github.com/mehDktb/Financial_portfolio_planing.git
cd Financial_portfolio_planing

Ensure you have required dependencies installed (Python, MiniZinc, pandas, sklearn):

pip install -r requirements.txt

To run the system, execute:

python3 main.py

Follow the steps outlined in the workflow to further interact with the system.

Contribution

Contributions and suggestions are welcome. Please submit pull requests or issues on GitHub. Contributions toward adding visualization and plotting capabilities are especially encouraged.

License

This project is licensed under the MIT License.