bhaskatripathi/MCP_Financial_Forecasting

Financial Forecasting MCP

A Model Context Protocol (MCP) server that provides SVM-based financial forecasting capabilities with OpenAI-powered interpretation and explanation.

Features

• Yahoo Finance Integration: Fetch real-time OHLC data for any stock symbol
• Advanced Feature Engineering: Technical indicators, price patterns, and lag features
• SVM Forecasting: A very basic Support Vector Machine for price prediction. More sophisticated models can be used for specialized implementations
• OpenAI Interpretation: AI-powered explanation and analysis of forecast results
• Hyperparameter Optimization: Automated parameter tuning for better performance
• Risk Assessment: Confidence metrics and risk factor identification

Tools Available

Data Operations

• fetch_stock_data: Get OHLC data from Yahoo Finance
• get_stock_info: Retrieve basic stock information
• validate_stock_symbol: Check if a symbol is valid and tradeable

Modeling & Forecasting

• train_svm_model: Train SVM model on historical data
• make_forecast: Generate price predictions with AI interpretation
• explain_model_performance: Get detailed model performance analysis

Installation

1. Clone or download the project

2. Install dependencies:

   pip install -e .
   

3. Set up environment variables: Create a .env file based on env.example:

   cp env.example .env
   

   Edit .env and add your OpenAI API key:

   OPENAI_API_KEY=your_openai_api_key_here
   

Configuration

The MCP server can be configured through environment variables:

• OPENAI_API_KEY: Required for forecast interpretation
• OPENAI_MODEL: OpenAI model to use (default: gpt-4o-mini)
• YAHOO_FINANCE_TIMEOUT: Timeout for Yahoo Finance requests (default: 30)
• SVM_KERNEL: Default SVM kernel (default: rbf)
• SVM_C: Default regularization parameter (default: 1.0)
• SVM_GAMMA: Default kernel coefficient (default: scale)
• FORECAST_DAYS: Default forecast horizon (default: 5)
• TRAIN_WINDOW_DAYS: Default training window (default: 252)

Usage Examples

Basic Stock Data Fetching

{
  "name": "fetch_stock_data",
  "arguments": {
    "symbol": "AAPL",
    "period": "1y"
  }
}

Training and Forecasting

{
  "name": "train_svm_model",
  "arguments": {
    "symbol": "AAPL",
    "forecast_days": 5,
    "optimize_params": true
  }
}

{
  "name": "make_forecast",
  "arguments": {
    "symbol": "AAPL",
    "include_interpretation": true,
    "market_context": "Recent earnings report shows strong growth"
  }
}

Technical Details

The idea is to use a very basic supervised learning model to demonstrated the concept.

SVM Model

• Kernels: Linear, Polynomial, RBF, Sigmoid
• Features: 40+ technical indicators and engineered features
• Validation: Time series cross-validation
• Metrics: R², RMSE, MAE, MAPE

Feature Engineering

• Technical Indicators: SMA, EMA, RSI, MACD, Bollinger Bands, Stochastic
• Price Features: Ratios, changes, log returns, ranges
• Lag Features: Historical price and volume lags
• Rolling Statistics: Moving averages and standard deviations

OpenAI Integration

• Model Performance Explanation: Detailed analysis of metrics and reliability
• Forecast Interpretation: Professional analysis with risk assessment
• Market Context: Integration of external market information
• Risk Factors: Identification of potential risks and limitations

Model Performance

The SVM models are evaluated using standard regression metrics:

• R² Score: Coefficient of determination (higher is better)
• RMSE: Root Mean Square Error (lower is better)
• MAE: Mean Absolute Error (lower is better)
• MAPE: Mean Absolute Percentage Error (lower is better)

Limitations and Disclaimers

⚠️ Important: This tool is for educational and research purposes only. It should not be used as the sole basis for investment decisions.

• Historical Performance: Past performance does not guarantee future results
• Market Volatility: Models may not capture extreme market events
• External Factors: News, earnings, and other events are not included in technical models
• Short-term Focus: SVM models work best for short to medium-term predictions
• Risk Management: Always use proper risk management and position sizing

Architecture

Financial Forecasting MCP
├── Data Layer (Yahoo Finance)
├── Feature Engineering (Technical Indicators)
├── ML Layer (SVM Models)
├── Interpretation Layer (OpenAI)
└── MCP Server (Tool Interface)

Development

Project Structure

src/financial_forecasting_mcp/
├── __init__.py
├── server.py              # Main MCP server
├── data_fetcher.py        # Yahoo Finance integration
├── feature_engineering.py # Technical indicators
├── svm_forecaster.py      # SVM modeling
└── openai_interpreter.py  # AI interpretation

Running Tests

pytest tests/

Code Quality

black src/
flake8 src/
mypy src/

License

This project is provided as-is for educational purposes. Please ensure you comply with Yahoo Finance's terms of service and OpenAI's usage policies.

Contributing

Contributions are welcome! Please ensure all code follows the existing style and includes appropriate tests.

Disclaimer: This software is for educational and research purposes only. It is not financial advice. Always consult with qualified financial professionals before making investment decisions.