😱 Someone just open sourced an AI hedge fund where Warren Buffett, Michael Burry, Charlie Munger & 9 other investing legends debate every stock and then a Portfolio Manager makes the final call.
No Bloomberg Terminal. No $25K brokerage minimums. No invitation-only fund.
It's called AI Hedge Fund.
Type a ticker.
Eighteen AI agents modeled on the greatest investors in history tear it apart from every angle. Value, growth, momentum, sentiment, fundamentals, technicals, risk. They argue. They signal.
The Portfolio Manager synthesizes it all into a final decision with position sizing included.
Not a stock screener.
Not a price alert tool.
A full multi-agent investment committee that thinks like the people who beat the market for decades.
No guesswork. No hot takes. No single point of failure.
Here's who's in the room when you analyze a stock:
→ Ben Graham Agent hunts for hidden gems trading below intrinsic value with a margin of safety and only buys when the numbers scream cheap
→ Michael Burry Agent goes full contrarian, digging for deep value the market has completely missed or misunderstood
→ Cathie Wood Agent bets on disruption and exponential growth, ignores short-term noise, and focuses on 5-year technology curves
→ Charlie Munger Agent refuses to touch anything that isn't a wonderful business at a fair price with a brutal filter that very few stocks pass
→ Stanley Druckenmiller Agent hunts macro asymmetry, positions where the upside is massive and the downside is contained
→ Phil Fisher Agent runs deep "scuttlebutt" research, the qualitative signals most analysts never bother to find
→ The Risk Manager sets hard position limits based on portfolio-level exposure before any order is generated
→ The Portfolio Manager synthesizes all 12 investor signals plus 4 quantitative agents into a single final trade decision with sizing
Here's how it actually works:
Each legendary investor agent reads the same financial data and outputs a directional signal: bullish, bearish, or neutral, with a confidence score and written reasoning in that investor's exact style. The Valuation Agent calculates intrinsic value. The Sentiment Agent reads the market mood.
The Fundamentals Agent checks the balance sheet. The Technicals Agent reads the chart.
The Risk Manager reviews all signals, sets position limits, and only then does the Portfolio Manager produce the final order. Every decision is traceable. You can see exactly which agents agreed, which dissented, and why.
Here's the wildest part:
You can also run the Backtester, feeding it a historical date range and watching how the 18-agent committee would have traded AAPL, MSFT, or NVDA over any period. Munger says hold. Burry says buy more. Wood says trim and rotate. You see the whole debate, dated, reasoned, and resolved across years of history.
poetry run python src/main.py --ticker AAPL,MSFT,NVDA
Works with GPT-4o, Claude, Gemini, DeepSeek, Groq, or local LLMs via Ollama. Free financial data for AAPL, GOOGL, MSFT, NVDA, and TSLA with no API key needed to start.
45.7K GitHub stars. 8K forks. 570 watchers. 799 commits. 32 contributors. Actively maintained.
100% Open Source. MIT License.
No Bloomberg Terminal. No $25K brokerage minimums. No invitation-only fund.
It's called AI Hedge Fund.
Type a ticker.
Eighteen AI agents modeled on the greatest investors in history tear it apart from every angle. Value, growth, momentum, sentiment, fundamentals, technicals, risk. They argue. They signal.
The Portfolio Manager synthesizes it all into a final decision with position sizing included.
Not a stock screener.
Not a price alert tool.
A full multi-agent investment committee that thinks like the people who beat the market for decades.
No guesswork. No hot takes. No single point of failure.
Here's who's in the room when you analyze a stock:
→ Ben Graham Agent hunts for hidden gems trading below intrinsic value with a margin of safety and only buys when the numbers scream cheap
→ Michael Burry Agent goes full contrarian, digging for deep value the market has completely missed or misunderstood
→ Cathie Wood Agent bets on disruption and exponential growth, ignores short-term noise, and focuses on 5-year technology curves
→ Charlie Munger Agent refuses to touch anything that isn't a wonderful business at a fair price with a brutal filter that very few stocks pass
→ Stanley Druckenmiller Agent hunts macro asymmetry, positions where the upside is massive and the downside is contained
→ Phil Fisher Agent runs deep "scuttlebutt" research, the qualitative signals most analysts never bother to find
→ The Risk Manager sets hard position limits based on portfolio-level exposure before any order is generated
→ The Portfolio Manager synthesizes all 12 investor signals plus 4 quantitative agents into a single final trade decision with sizing
Here's how it actually works:
Each legendary investor agent reads the same financial data and outputs a directional signal: bullish, bearish, or neutral, with a confidence score and written reasoning in that investor's exact style. The Valuation Agent calculates intrinsic value. The Sentiment Agent reads the market mood.
The Fundamentals Agent checks the balance sheet. The Technicals Agent reads the chart.
The Risk Manager reviews all signals, sets position limits, and only then does the Portfolio Manager produce the final order. Every decision is traceable. You can see exactly which agents agreed, which dissented, and why.
Here's the wildest part:
You can also run the Backtester, feeding it a historical date range and watching how the 18-agent committee would have traded AAPL, MSFT, or NVDA over any period. Munger says hold. Burry says buy more. Wood says trim and rotate. You see the whole debate, dated, reasoned, and resolved across years of history.
poetry run python src/main.py --ticker AAPL,MSFT,NVDA
Works with GPT-4o, Claude, Gemini, DeepSeek, Groq, or local LLMs via Ollama. Free financial data for AAPL, GOOGL, MSFT, NVDA, and TSLA with no API key needed to start.
45.7K GitHub stars. 8K forks. 570 watchers. 799 commits. 32 contributors. Actively maintained.
100% Open Source. MIT License.
Here is the repo🔗 github.com/virattt/ai-hed…
Thank you for reading this.
If you enjoyed this post:
1. Follow me @KanikaBK for more of these
2. RT the tweet below to share this thread with your audience
If you enjoyed this post:
1. Follow me @KanikaBK for more of these
2. RT the tweet below to share this thread with your audience
https://twitter.com/1483409796531625987/status/2034637287729115617
• • •
Missing some Tweet in this thread? You can try to
force a refresh
