This project implements a heuristic algorithm for financial portfolio optimization. It utilizes a Biased-Randomized (BR) approach combined with Local Search to find optimal asset allocations, aiming to maximize the Sharpe Ratio and expected returns while managing risk. It is accesible at portfolio-br.streamlit.app.

The application features an interactive web interface built with Streamlit, allowing users to configure parameters, select assets, and compare the custom algorithm against industry-standard benchmarks like Markowitz Mean-Variance, Risk Parity, and Hierarchical Risk Parity (HRP).

• Biased-Randomized Algorithm: Implements a multi-start heuristic using geometric and triangular distributions to explore the solution space efficiently.
• Local Search Optimization: Refines solutions generated by the constructive phase to avoid local optima.
• Flexible Objective Function: Allows users to weight the optimization goal between risk-adjusted returns (Sharpe Ratio) and raw returns.
• Benchmark Comparison: Automatically compares results against:
  • Equal Weight (1/N)
  • Minimum Variance
  • Risk Parity
  • Hierarchical Risk Parity (HRP)
  • Maximum Diversification
  • Greedy Heuristic
• Interactive Visualizations: Includes efficient frontier plots, allocation charts, rolling performance metrics, and correlation heatmaps using Plotly.
• Data Management: Fetches real-time data via Yahoo Finance and supports caching/saving datasets for offline analysis.

The core algorithm departs from traditional quadratic programming by using a biased-randomized heuristic approach. It constructs portfolios iteratively:

1. Ranking: Assets are ranked based on individual performance metrics (e.g., Sharpe Ratio).

2. Biased Selection: Assets are selected for the portfolio using a non-uniform distribution (Geometric or Triangular), favoring higher-ranked assets but maintaining randomness to ensure diversity.

3. Allocation: Weights are assigned based on a randomized allocation scheme constrained by lower and upper bounds.

4. Refinement: A local search process adjusts weights to maximize the objective function defined as:

   $f(w) = \gamma * SharpeRatio + \omega * PortfolioReturn$

The codebase is modularized into three main components to ensure separation of concerns:

• Portfolio Optimizer.py: The entry point for the Streamlit application. Handles user interface, state management, and interaction flow.
• financial_model.py: Contains the core business logic, including data fetching, mathematical calculations, and the implementation of the optimization algorithms.
• visualization.py: Dedicated module for generating Plotly charts and graphs, keeping the UI logic clean.

1. Clone the repository:

   git clone https://github.com/hsannav/portfolio-deployment.git
   cd portfolio-deployment

2. Create a virtual environment (recommended):

   python -m venv venv
   source venv/bin/activate  # On Windows use `venv\Scripts\activate`

3. Install the required dependencies:

   pip install -r requirements.txt

1. Run the Streamlit application:

   streamlit run 'Portfolio Optimizer.py'

2. The application will open in your default web browser.

3. In the sidebar:

   • Select "New/Existing configuration".
   • Choose a preset portfolio (e.g., Tech Giants) or enter custom tickers.
   • Adjust the objective weights (Sharpe vs. Return).
   • Set optimization parameters (number of iterations, beta).
   • Click "Run optimization".

• Python 3.8+
• streamlit
• numpy
• pandas
• yfinance
• scipy
• plotly

This project is for educational and demonstration purposes only. It does not constitute financial advice. Algorithmic trading and portfolio optimization involve significant risk.