Hey HN — I'm the author. Let me give a proper intro since the original title didn't do a great job explaining what this actually does.
HaiInvestor lets you pick any stock ticker and watch 6 AI personas — each modeled after a legendary investor — debate it in real time:
Warren Buffett (value / moat-focused)
Charlie Munger (mental models / inversion)
Michael Burry (contrarian / macro risk)
Peter Lynch (growth at reasonable price)
Cathie Wood (disruptive tech / long arc)
Bill Ackman (activist / concentrated bets)
Each agent reads live news headlines + financial data, then outputs a BULLISH / BEARISH / NEUTRAL signal with a confidence score and reasoning in that investor's voice. A Portfolio Manager agent then synthesizes a final BUY / SELL / HOLD verdict.
There are three modes:
- Single Stock — debate on one ticker
- Portfolio — analyze your whole holdings mix
- Backtest — test a strategy over historical data
A few honest caveats: this is purely educational/fun, not financial advice. The agents reason from news only (no full fundamentals yet), and LLMs can hallucinate. I'd love feedback on the multi-agent architecture, what data sources to add next, or any other ideas. Happy to answer questions!
HaiInvestor lets you pick any stock ticker and watch 6 AI personas — each modeled after a legendary investor — debate it in real time:
Each agent reads live news headlines + financial data, then outputs a BULLISH / BEARISH / NEUTRAL signal with a confidence score and reasoning in that investor's voice. A Portfolio Manager agent then synthesizes a final BUY / SELL / HOLD verdict.There are three modes: - Single Stock — debate on one ticker - Portfolio — analyze your whole holdings mix - Backtest — test a strategy over historical data
Tech stack: Python, Streamlit, OpenAI GPT-4o-mini, yfinance, BeautifulSoup.
A few honest caveats: this is purely educational/fun, not financial advice. The agents reason from news only (no full fundamentals yet), and LLMs can hallucinate. I'd love feedback on the multi-agent architecture, what data sources to add next, or any other ideas. Happy to answer questions!