An open-source quantitative analysis engine and multi-LLM ranking framework for Indian Mutual Funds (Small Cap, Mid Cap, Total Market, Multi Asset).

📊 View Live Ranks (Google Sheet) | 📖 Methodology | 🚀 Quickstart

MF Screener AI evaluates 200+ Indian mutual fund schemes using historical NAV data from TickerTape. Instead of relying on a single fixed formula, we act as an "LLM Arena": we task Claude, Gemini, GPT, and Grok to independently design quantitative scoring strategies.

The models evaluate funds across 30+ risk, return, and portfolio metrics—including Sharpe, Sortino, Alpha, Beta, Max Drawdown, and verifiable SIP XIRR. The final result is a normalized, composite ranking of funds based on the consensus of all four AI models.

Most mutual fund screeners provide basic star ratings based on past CAGR. We built this to:

1. Provide transparent, open-source quantitative screening.
2. Give investors actionable insights matching real-world investing scenarios (e.g., heavily weighting historical SIP XIRR rather than just point-to-point returns).
3. Test if diverse AI reasoning models can identify resilient, high-quality funds better than standard static filters.
4. Verifiable Performance: We compute the rolling SIP XIRR for every strategy against standard NIFTY benchmarks to prove the real-world value of the LLM rankings. You can view the Latest Strategy XIRR in the rankings.

The screening pipeline runs reliably fully automated via GitHub Actions every week.

1. Data Fetching: Pulls the latest mutual fund metadata and historical NAV data.
2. Algorithm Scoring: Python scripts written by Claude, Gemini, GPT, and Grok analyze the data.
3. Compilation: Scores are combined, normalized, and mapped to a final composite ranking.
4. Publishing: Results are updated in the Live Google Sheet and static indexable rankings.

This project is designed to be forked and customized. You can easily add your own quantitative logic or ask an AI to generate one for a specific sector.

1. Fork this repository.
2. Modify tasks/2_algorithm.md to specify your target sector and metrics.
3. Generate a new scoring algorithm.
4. Run the script to see how your strategy performs. See the Setup Guide for detailed instructions.

• Not Financial Advice: The strategies, metrics, and scores generated in this project are for educational and experimental purposes only. They should not be considered as financial advice.
• AI-Generated Models: The quantitative scoring scripts were generated by AI models. Always verify the logic (e.g., checking the SIP XIRR calculations).
• Data Accuracy: Data is fetched from third-party sources (TickerTape); accuracy is not guaranteed.

What is the best open-source mutual fund screener for India? MF Screener AI provides an open-source alternative to commercial screeners, focusing on quantitative metrics and AI-driven composite rankings for Indian mutual funds.

Does it filter by Sharpe Ratio, Sortino Ratio, and Alpha? Yes, the underlying AI models evaluate funds across 30+ metrics including Sharpe, Sortino, Jensen's Alpha, Beta, Downside Capture, and Max Drawdown.

Does it support Small Cap, Mid Cap, and Multi Asset funds? Yes, it actively tracks and ranks 200+ schemes across Small Cap, Mid Cap, Total Market, and Multi Asset categories.

How does it calculate SIP returns? The models calculate rolling SIP XIRR based on historical weekly NAV data, simulating real retail investor outcomes rather than just point-to-point CAGR.