AI-Powered Stock Analysis with Multi-Agent Collaboration
Features • Quick Start • Architecture • Usage • Configuration
TradingMind is a multi-agent framework that mirrors real-world trading firms. Specialized LLM-powered agents—analysts, researchers, traders, and risk managers—collaborate through debate to evaluate market conditions and inform trading decisions.
Disclaimer: For research purposes only. Not financial advice.
| Category | Description |
|---|---|
| Multi-Agent Analysis | 4 specialized analysts (Technical, Sentiment, News, Fundamentals) |
| Bull vs Bear Debate | Researchers argue opposing viewpoints for balanced insights |
| Risk Management | 3-way risk debate (Aggressive, Conservative, Neutral) |
| Memory & Learning | ChromaDB-powered system learns from past decisions |
| Multiple LLMs | OpenAI, Anthropic Claude, DeepSeek, Google |
| Web UI + CLI | Modern React frontend & rich terminal interface |
# 1. Install uv (fast Python package manager)
curl -LsSf https://astral.sh/uv/install.sh | sh
# 2. Clone and install
git clone https://github.com/your-repo/tradingmind.git
cd tradingmind
uv sync
# 3. Configure API keys
cp .env.example .env
# Edit .env with your API keys (see Configuration section)
# 4. Run CLI
uv run python -m cli.main
# Or start Web UI (requires two terminals)
uv run python api/main.py # Terminal 1: Backend
cd frontend && npm i && npm run dev # Terminal 2: Frontend
# Open http://localhost:3000User Input (Ticker, Date)
│
▼
┌─────────────────────────────────────────────────────────┐
│ ANALYST TEAM │
│ ┌──────────┐ ┌──────────┐ ┌──────────┐ ┌──────────┐ │
│ │Technical │ │Sentiment │ │ News │ │Fundamental│ │
│ │ Analyst │ │ Analyst │ │ Analyst │ │ Analyst │ │
│ └────┬─────┘ └────┬─────┘ └────┬─────┘ └────┬──────┘ │
└───────┼────────────┼────────────┼────────────┼──────────┘
└────────────┴─────┬──────┴────────────┘
▼
┌─────────────────────────────────────────────────────────┐
│ RESEARCHER TEAM │
│ ┌──────────────┐ ┌──────────────┐ │
│ │ Bull │◄──►│ Bear │ │
│ │ Researcher │ │ Researcher │ │
│ └──────┬───────┘ └──────┬───────┘ │
│ └───────┬───────────┘ │
│ DEBATE │
└────────────────────────┼────────────────────────────────┘
▼
┌─────────────────────────────────────────────────────────┐
│ TRADER AGENT │
│ Synthesizes reports → BUY / SELL / HOLD │
└────────────────────────┼────────────────────────────────┘
▼
┌─────────────────────────────────────────────────────────┐
│ RISK MANAGEMENT TEAM │
│ ┌──────────┐ ┌──────────┐ ┌──────────┐ │
│ │Aggressive│ │ Neutral │ │Conservative│ │
│ └────┬─────┘ └────┬─────┘ └────┬──────┘ │
│ └──────────────┼──────────────┘ │
│ RISK DEBATE │
└────────────────────────┼────────────────────────────────┘
▼
┌─────────────────────────────────────────────────────────┐
│ PORTFOLIO MANAGER │
│ Final Decision + Reasoning │
└─────────────────────────────────────────────────────────┘
| Team | Agent | Responsibility |
|---|---|---|
| Analysts | Technical | MACD, RSI, Bollinger Bands analysis |
| Sentiment | Social media & public sentiment | |
| News | Global news & macro indicators | |
| Fundamentals | Financials, earnings, balance sheets | |
| Researchers | Bull | Advocates bullish positions with evidence |
| Bear | Identifies risks and bearish factors | |
| Trading | Trader | Synthesizes reports, proposes trades |
| Risk | Risk Team | 3-way debate on risk tolerance |
| Portfolio Mgr | Final approval/rejection |
# Terminal 1: Backend
uv run python api/main.py
# Terminal 2: Frontend
cd frontend && npm run dev| URL | Description |
|---|---|
| http://localhost:3000 | Web UI |
| http://localhost:8001 | API |
| http://localhost:8001/docs | Swagger Docs |
uv run python -m cli.mainThe CLI prompts for ticker, date, and LLM provider.
from backend.graph.trading_graph import TradingMindGraph
graph = TradingMindGraph()
final_state, decision = graph.propagate("NVDA", "2024-12-01")
print(f"Decision: {decision}")Create .env from template:
cp .env.example .env| Variable | Required | Description |
|---|---|---|
OPENAI_API_KEY |
Yes* | OpenAI API key |
ANTHROPIC_API_KEY |
Yes* | Anthropic API key |
DEEPSEEK_API_KEY |
Yes* | DeepSeek API key (budget option) |
ALPHA_VANTAGE_API_KEY |
Yes | Market data (free key) |
USE_MEMORY |
No | Enable learning system (default: true) |
REDIS_HOST |
No | Redis host for caching |
DEBUG_LOGGING |
No | Enable verbose logs |
*At least one LLM provider required
| Provider | Get API Key | Notes |
|---|---|---|
| OpenAI | platform.openai.com | GPT-4o recommended |
| Anthropic | console.anthropic.com | Claude 3.5 Sonnet |
| DeepSeek | platform.deepseek.com | Budget-friendly |
tradingmind/
├── api/ # FastAPI backend
├── backend/
│ ├── agents/ # All agent implementations
│ │ ├── analysts/ # Market, News, Social, Fundamentals
│ │ ├── researchers/ # Bull and Bear
│ │ ├── trader/ # Trading agent
│ │ └── risk_debate/ # Risk analysts
│ ├── graph/ # LangGraph workflow
│ ├── dataflows/ # Data vendor integrations
│ └── analysis/ # Risk & position sizing
├── cli/ # Terminal interface
├── frontend/ # React + Tailwind UI
└── docs/ # Documentation & images
- Agent Framework: LangChain + LangGraph
- LLM Providers: OpenAI, Anthropic, DeepSeek, Google
- Data Sources: yfinance, Alpha Vantage, Finnhub, SEC EDGAR
- Memory: ChromaDB (vector storage)
- Backend: FastAPI + Redis
- Frontend: React + Vite + Tailwind CSS
MIT License - See LICENSE for details.