What It Does
The AI Hedge Fund Agent turns a plain-language question—"Should I buy Nvidia?"—into a fully-formed trading decision. In short, it's a pocket-sized quant desk that speaks natural language.
Features:
- 🧠 Enhanced LLM Query Parsing - Single smart call extracts portfolio context, risk signals, and trade intent
- ⚡ Real-Time Market Data - Live data from stocks, crypto, and ETFs via yfinance & CoinGecko APIs
- 📊 Multi-Modal Analysis - Technical indicators (EMA 8/21, RSI, MACD, volume) + sentiment analysis + real-time news
- 🎯 Intelligent Routing - LangGraph workflow adapts based on query complexity and user intent
- 💼 Portfolio Decision Engine - Makes concrete BUY/SELL/HOLD decisions with specific quantities
- 🛡️ Risk Management - Comprehensive risk assessment with position sizing recommendations
- 📋 Actionable Output - Concrete trading tables with entry/exit prices, stop losses, and position sizes
- 🚀 Agentuity Native - Built for seamless deployment on Agentuity platform
How It Works
The workflow is a six-node LangGraph pipeline:
- Query Parsing – extracts trade intent, existing positions, risk tolerance.
- Asset Search – validates tickers & fetches live quotes.
- Trade Analysis – merges technical + sentiment signals.
- Risk Assessment – gauges volatility, concentration, and capital at risk.
- Portfolio Manager – decides BUY/SELL/HOLD, position size, targets.
- Response Generator – formats everything into sleek tables or insights.
Real-Time Analysis Highlights
- Technical: EMA crossovers, RSI (14), MACD histogram, volume spikes.
- Sentiment: News API headlines, CNN Fear/Greed, VIX, FRED macro feeds.
- AI Insights: Correlation shocks, event calendars, emerging themes.
If trade intent ≥ 70 %, the agent surfaces a concrete order ticket. Otherwise you get research commentary only.
Example Actionable Output
High Trade Intent Query: "Should I buy Apple stock?"
Medium Trade Intent Query: "How is Apple looking?"
Some Notable Things
- Agent-per-Node Design – Each LangGraph node runs as its own Agentuity agent, so you can scale risk calculations separately from, say, news scraping.
- Single-Call Parsing – One LLM invocation tags intent, asset list, portfolio, and risk appetite—reducing latency + cost.
- Pro-Grade Risk Management – Position sizing obeys volatility, cash balance, and user-defined limits; prompts warn about concentration.
- Instant Cloud Deploy – Because it's Agentuity-native, a two-command flow (
agentuity project import
→agentuity deploy
) gets you an HTTPS endpoint in minutes.
How to Use It
You can deploy it on Agentuity in one click:
Or you can clone the repo and deploy it yourself:
Check the README for details on env vars and the like you'll need. To make it your own on Agentuity, it's as simple as:
-
Clone the repo
-
agentuity project import
-
agentuity deploy
Example Queries
- "I own 100 shares of TSLA—should I sell half?"
- "Is Bitcoin overbought today?"
- "What's my portfolio's overall risk level?"
Community Spotlight
Have a cool agent? Fork this template, deploy on Agentuity, and share your project in our Discord. We might feature you next!