📈 AI-Powered Investment Portfolio Optimization Agent

An advanced financial agent built using LangGraph, LangChain, and Streamlit that automates the entire portfolio management lifecycle—from user profiling and ticker selection to risk-adjusted optimization and real-time market sentiment analysis.

🚀 Key Features

1. 🤖 AI-Driven User Profiling

• Infers your investment profile (risk appetite and horizon) from natural language descriptions.
• Preserves explicit user constraints while adding AI-powered insights.

2. 🔍 Dynamic Ticker Selection

• Recommends relevant stocks, ETFs, and sectors based on your preferences.
• Uses LLM reasoning to filter tickers that align with current market trends and your risk profile.

3. 📊 Advanced Portfolio Optimization

• Max Sharpe Ratio (Tangency Portfolio): Automatically identifies the optimal risk-return trade-off.
• Stable Dynamic Leverage: Implements a prudent Equity Multiplier (up to 2.5x) to enhance returns without degrading portfolio efficiency.
• Monte Carlo Simulations: Runs thousands of simulations to calculate 95% Value-at-Risk (VaR) and Expected Shortfall.

4. 📰 Real-Time Market Intelligence

• Sentiment Analysis: Fetches and analyzes news headlines from Finnhub & Yahoo Finance.
• Fundamental Analysis: Pulls P/E ratios, Dividend Yields, Beta, and other key financial ratios.
• Technical Indicators: Generates SMA, RSI, and momentum charts.

5. 🏗️ Agentic Workflow (LangGraph)

• A multi-node state machine orchestrates the process:
  • fetch_market_data ➡️ infer_user_profile ➡️ select_tickers ➡️ analyze_market ➡️ optimize_portfolio ➡️ assess_portfolio.

🛠️ Tech Stack

• Framework: LangGraph, LangChain
• LLM: OpenAI (GPT-4o)
• UI: Streamlit
• Financial APIs: Finnhub, yfinance
• Optimization: CVXPY, NumPy, Pandas
• Visualization: Plotly, Altair

⚙️ Setup & Installation

1. Clone the repository:

   git clone <repository-url>
   cd "Investment Portfolio Optimization Agent using LangGraph"

2. Install dependencies:

   pip install -r requirements.txt

3. Configure Environment Variables:
   Create a .env file or set up Streamlit secrets:

   OPENAI_API_KEY=your_openai_key
   FINNHUB_API_KEY=your_finnhub_key

4. Run the Application:

   streamlit run app.py

📈 Architecture

The system uses a State Graph approach where each node modifies a shared AgentState object. This ensures traceability and allows for complex conditional logic (e.g., rebalancing versus hedging based on market sentiment).

🛡️ Risk Management

• Leverage Caps: Strictly enforced limits to prevent over-exposure (Aggressive: 2.5x max).
• Metric Sanitization: Proper handling of percentage-based metrics (Dividend Yield, Beta) to ensure realistic projections.
• Action Recommendations: Context-aware suggestions (Hold, Rebalance, Hedge) based on portfolio-wide risk scoring.