Nifty LLM Research

A research project using LLMs to analyze and forecast NSE-listed stocks, with a focus on portfolio optimization.

Overview

This project uses Google's Gemini AI to:

1. Analyze NSE-listed stocks and generate price forecasts
2. Optimize stock portfolios based on these forecasts
3. Track portfolio performance

Key Features

• Stock Analysis: Deep research on individual stocks using Google Search integration
• Portfolio Optimization: Selection of optimal stock combinations based on forecasts
• Performance Tracking: Historical tracking of portfolio performance
• Output Generation: Detailed reports in both JSON and Markdown formats

Getting Started

Prerequisites

• Python 3.9+
• MongoDB
• Google Cloud API key for Gemini AI
• Google Search API key

Installation

1. Clone the repository:

git clone https://github.com/yourusername/nifty-llm-research.git
cd nifty-llm-research

2. Install dependencies:

pip install -r requirements.txt

3. Set up environment variables:

cp env.template .env
# Edit .env with your API keys and configuration

Usage

1. Seed the database with default prompts:

python scripts/seed_prompts.py

2. Generate stock forecasts:

python scripts/analyze_stocks.py --index "NIFTY 50" --force-nse

3. Generate portfolio recommendations:

python scripts/generate_portfolio.py --index "NIFTY 50" --filter-top-n 20 --basket-size-k 5

Understanding the Code

Development History

This project was developed using Cursor IDE's AI pair programming capabilities. To see how the code was written:

1. Check the prompts/ directory to see the actual Cursor prompts used during development
2. Each prompt file shows the exact conversation that led to the implementation
3. This serves as a proof of authenticity and helps understand the development process

Prompts

The core of this project lies in its carefully crafted prompts. To understand how the code works:

1. Check the scripts/seed_prompts.py file to see the default prompts
2. Review the prompt configurations in the database
3. Understand how each prompt is used in the respective agents

Outputs

All portfolio recommendations are saved in the docs/baskets directory:

• JSON files containing the full analysis
• Markdown files with formatted tables and explanations
• Files are named as: {index}-{timestamp}-{n}-{k}.{json/md}

Project Structure

nifty-llm-research/
├── config/             # Configuration files
├── docs/              # Documentation and outputs
│   └── baskets/       # Portfolio recommendations
├── prompts/           # Development history and Cursor prompts
├── scripts/           # Command-line scripts
├── src/              # Source code
│   ├── agents/       # LLM agents
│   ├── db/          # Database models and utilities
│   ├── utils/       # Utility functions
│   └── visualization/# Plotting and visualization
└── tests/            # Test files

Contributing

1. Fork the repository
2. Create a feature branch
3. Commit your changes
4. Push to the branch
5. Create a Pull Request

License

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

Acknowledgments

• Google Gemini AI for the LLM capabilities
• NSE for stock data
• Contributors and maintainers