AI-powered multi-agent trading analysis system with real-time dashboard.

cd backend
source ../venv/bin/activate
python demo_server.py

cd ai-trader-collab
npm install  # First time only
npm run dev

Visit: http://localhost:8080

Click "Start Analysis" with ticker "AAPL" to see the demo!

• Market Analyst - Technical indicators (RSI, MACD, Bollinger Bands, Moving Averages)
• News Analyst - Sentiment analysis from news sources
• Fundamentals Analyst - Financial metrics (P/E, revenue, earnings, cash flow)
• Payment Flow Analyst - Transaction patterns and payment ecosystem health
• Risk Manager - Risk assessment and position sizing
• Trader - Final trading decision synthesis

• Clean, conversational agent messages
• Interactive Plotly charts (4 visualizations)
• Expandable detailed reports
• Trading decisions with confidence scores
• WebSocket + polling for reliable updates

Demo Mode (demo_server.py)

• Hardcoded AAPL analysis based on real data
• Perfect for presentations
• No API credits needed
• All 6 agents with professional analysis

Live Mode (api_server.py)

• Real AI-powered analysis using Claude
• Requires Anthropic API credits
• LLM-powered message summarization
• Dynamic analysis for any ticker

Backend:

• Python 3.13
• FastAPI (REST + WebSocket)
• LangGraph (Agent orchestration)
• Anthropic Claude (LLM)
• TradingAgents library

Frontend:

• React + TypeScript
• Vite
• shadcn/ui components
• Tailwind CSS
• Plotly charts

ΣIGMA/
├── backend/
│   ├── demo_server.py              # Demo mode server
│   ├── api_server.py               # Live analysis server  
│   ├── requirements.txt            # Python dependencies
│   ├── enhanced_visualization_*.html  # Chart files (4)
│   └── tradingagents/              # Core agent library
│       ├── agents/                 # All agent implementations
│       ├── graph/                  # LangGraph workflow
│       └── dataflows/              # Data providers
│
├── ai-trader-collab/              # React frontend
│   ├── src/
│   │   ├── components/            # UI components
│   │   └── hooks/                 # React hooks
│   └── package.json
│
└── venv/                          # Python virtual environment

Backend Environment Variables:

ANTHROPIC_API_KEY=your_api_key_here
DEMO_MODE=true  # or false for live mode

Frontend:

• Connects to http://localhost:8002 (backend)
• Runs on http://localhost:8080 (configurable in vite.config.ts)

• POST /api/analyze - Start analysis
• GET /api/analysis/{id}/messages - Get agent messages (polling)
• GET /api/analysis/{id}/reports - Get detailed reports
• GET /api/charts/{id} - Get chart list
• GET /api/chart/{filename} - Serve chart HTML
• WS /ws/analysis/{id} - WebSocket for real-time updates

1. Start backend (demo or live mode)
2. Start frontend
3. Enter ticker symbol (e.g., AAPL, TSLA, MSFT)
4. Click "Start Analysis"
5. Watch agents analyze in real-time
6. View charts and reports
7. See final trading decision

# Backend setup
cd backend
python3 -m venv ../venv
source ../venv/bin/activate
pip install -r requirements.txt

# Frontend setup
cd ../ai-trader-collab
npm install

The demo mode showcases real AAPL analysis data:

• RSI: 71.05 (overbought)
• Price: $258.02
• Revenue: $408.6B (9.6% growth)
• 96M monthly transactions
• Final Decision: SELL (75% confidence)

Perfect for presentations and showcasing capabilities!

Aryan Sinha
📧 sinha.arya@northeastern.edu
🎓 Northeastern University

Shourya Dewansh
🎓 Wentworth Institute of Technology

Arzu Malkoch
🎓 Northeastern University

This project was developed as part of a trading analysis system showcasing multi-agent AI collaboration. The system integrates multiple specialized AI agents for comprehensive stock analysis with real-time visualization.

• Custom multi-agent architecture using LangGraph
• Anthropic Claude AI for agent intelligence
• React + TypeScript frontend
• FastAPI backend with WebSocket support

See LICENSE file for details.