xbtlin/ai-berkshire

Achieved over 1.46 million RMB in real portfolio returns over two years, significantly outperforming the S&P 500. AI Berkshire is a comprehensive, open-source value investing research framework built on top of Claude Code, systematically digitizing and replicating the investment methodologies of Warren Buffett, Charlie Munger, Duan Yongping, and Li Lu to make professional-grade research accessible.

🔹 **Definitive Decision Making**: Unlike typical AI models that offer balanced but ultimately useless advice, this framework forces definitive 'pass', 'fail', or 'grey area' ratings. It provides concrete valuation ranges, price targets, and multi-layer action items for clear decision-making.

🔹 **Multi-Agent Adversarial Analysis**: Initiates four parallel AI agents simulating different masters' perspectives. It pits views on business models, financial valuation, reverse thinking, and long-term moat against each other, highlighting crucial trade-offs and eliminating cognitive biases.

🔹 **Financial Calculation Rigor**: Integrates python decimal-based calculation scripts instead of unreliable model-based floats. This ensures that market caps, stock prices, and valuation ratios from multiple independent sources are rigorously verified and matched.

🔹 **Standardized Research Pipeline**: Guarantees identical analysis depth, structure, and checklist checkpoints for every target company. This consistency makes it seamless to perform side-by-side comparisons of different tickers or track the same business's evolution over long periods.

https://github.com/xbtlin/ai-berkshire