Autonomous Arbitrage & Leverage Trading Agent

Crypto-Stocks is an autonomous trading agent system designed to identify and capitalize on arbitrage opportunities across crypto exchanges and stock data providers. The system uses real-time price monitoring, leverage calculations, and AI-driven sentiment analysis to target 20-30% returns.

Core Strategies:

• Cross-Exchange Arbitrage: Price gaps between exchanges
• Leveraged Trading: 5x, 10x, 20x multipliers on high-probability opportunities
• Risk Management: Kill switch, margin monitoring, PDT compliance
• AI Sentiment: Long-term analysis using Gemini 2.0 Flash

Stack: TypeScript, Node.js, CCXT, Drizzle ORM, PostgreSQL (Railway), OpenRouter

---

🏗️ Project Structure

/agent
├── src/
│   ├── agents/
│   │   ├── crypto/
│   │   │   ├── arbitrage.ts    # WebSocket listener (CCXT Pro)
│   │   │   ├── leverage.ts     # ROI calculator (5x, 10x, 20x)
│   │   │   └── index.ts        # Crypto agents runner
│   │   └── stocks/
│   │       ├── arbitrage.ts    # Price gap scanner (Alpha Vantage/Polygon)
│   │       ├── leverage.ts     # Margin/PDT rule tracker
│   │       ├── long-term.ts    # AI sentiment analysis
│   │       └── index.ts        # Stock agents runner
│   ├── database/
│   │   └── migrations.sql      # SQL schema for Railway
│   ├── db/
│   │   ├── schema.ts           # Drizzle ORM schema
│   │   └── index.ts            # Database connection
│   └── index.ts                # Main entry point
├── package.json
└── drizzle.config.ts

---

📊 Database Schema

The system uses 4 core tables on Railway PostgreSQL:

1. market_depth

Real-time pricing data from exchanges and data providers

• Exchange name, symbol, best bid/ask, timestamp
• Indexes on symbol/timestamp for fast queries

2. leverage_opportunities

Detected arbitrage opportunities with leverage calculations

• Asset type (crypto/stock), price gap %, leverage multiplier
• Projected ROI after fees, status tracking

3. risk_management

Kill switch and performance tracking (singleton table)

• Consecutive losses counter
• Total P/L, win rate, kill switch status
• Auto-activates after 3 losses in 24 hours

4. execution_log

History of all simulated/real trades

• Buy/sell actions, fees paid, profit/loss
• Links to opportunities for backtesting

---

🚀 Getting Started

1. Prerequisites

• Node.js 18+ (or use Docker)
• Railway PostgreSQL database
• API Keys (optional but recommended):
  • OpenRouter (for AI sentiment analysis)
  • Alpha Vantage (for stock prices)
  • Polygon.io (for stock prices)

2. Installation

cd agent
npm install

3. Environment Setup

Copy .env.example to .env and configure:

# Database (Railway)
DATABASE_URL="postgresql://user:pass@host:port/dbname"

# Optional: Stock Data APIs
ALPHA_VANTAGE_API_KEY="your_key_here"
POLYGON_API_KEY="your_key_here"

# Optional: AI Analysis
OPENROUTER_API_KEY="your_key_here"

# Agent Control (default: both enabled)
RUN_CRYPTO=true
RUN_STOCKS=true

4. Database Migration

Run the SQL migration on Railway:

npm run db:migrate

Or manually run agent/src/database/migrations.sql in Railway's PostgreSQL console.

5. Run the Agents

Run all agents:

npm run dev

Run crypto only:

npm run run:crypto

Run stocks only:

npm run run:stocks

---

🎯 How It Works

Crypto Agents

Arbitrage Agent:

• Monitors BTC/USDT on Coinbase & Kraken via WebSocket
• Detects price gaps between exchanges
• Logs data to market_depth table every tick

Leverage Agent:

• Scans market_depth every 10 seconds
• Calculates ROI with 5x, 10x, 20x leverage
• Logs opportunities hitting 20-30% target (after 0.2% fees)
• Respects kill switch (3 losses in 24h)

Stock Agents

Arbitrage Agent:

• Polls Alpha Vantage & Polygon every 60 seconds
• Detects price gaps between data providers
• Logs data for top stocks (AAPL, MSFT, GOOGL, TSLA, NVDA)

Leverage Agent:

• Scans for stock arbitrage opportunities
• Enforces PDT rules (Pattern Day Trader)
  • Accounts <$25k: Max 3 day trades per 5 days, 2x leverage
  • Accounts ≥$25k: Unlimited day trades, 4x leverage
• Calculates ROI with available leverage

Long-Term Agent:

• AI-driven sentiment analysis (Gemini 2.0 Flash)
• Analyzes market trends, fundamentals, technical indicators
• Provides buy/sell/hold recommendations
• Runs every hour for configured symbols

---

⚠️ Risk Management

Kill Switch

Automatically triggered when:

• 3 consecutive losses within 24 hours
• Prevents further opportunity logging
• Must be manually reset after review

Margin Protection

• Real-time liquidation price monitoring (planned)
• Slippage checks before execution
• Position sizing based on account balance

Simulated Trading

All trades are simulated by default for safety:

• Logs to execution_log table
• No real money at risk
• Perfect for backtesting and validation

---

📈 Deployment (Raspberry Pi)

1. Setup on Raspberry Pi

# Clone the repository
git clone git@github.com:ChrisForti/crypto-stocks.git
cd crypto-stocks/agent

# Install dependencies
npm install

# Configure environment
cp .env.example .env
nano .env  # Add your DATABASE_URL and API keys

2. Run Database Migrations

npm run db:migrate

3. Start with PM2 (Background Process)

# Install PM2 globally
npm install -g pm2

# Start all agents
pm2 start npm --name "arbitrage-agent" -- run start

# Or use tsx for development
pm2 start tsx --name "arbitrage-agent" -- src/index.ts

# View logs
pm2 logs arbitrage-agent

# Monitor status
pm2 status

# Restart
pm2 restart arbitrage-agent

4. Auto-Start on Boot

# Save PM2 configuration
pm2 save

# Setup startup script
pm2 startup
# Follow the instructions printed

---

🔧 Configuration Options

Environment Variables

Variable	Description	Required	Default
DATABASE_URL	Railway PostgreSQL connection string	✅ Yes	-
RUN_CRYPTO	Enable crypto agents	❌ No	true
RUN_STOCKS	Enable stock agents	❌ No	true
OPENROUTER_API_KEY	OpenRouter API for AI analysis	❌ No	-
ALPHA_VANTAGE_API_KEY	Alpha Vantage for stock prices	❌ No	-
POLYGON_API_KEY	Polygon.io for stock prices	❌ No	-

---

📊 Monitoring & Analytics

View Statistics

The agents log all opportunities to the database. Query them:

-- View recent opportunities
SELECT * FROM leverage_opportunities
WHERE projected_roi_pct >= 20
ORDER BY created_at DESC
LIMIT 20;

-- Check kill switch status
SELECT * FROM risk_management;

-- View win rate
SELECT
  asset_type,
  COUNT(*) as total_trades,
  SUM(CASE WHEN profit_loss > 0 THEN 1 ELSE 0 END) as wins,
  AVG(profit_loss) as avg_profit
FROM execution_log
GROUP BY asset_type;

Export for AI Audit

After 48 hours of simulation:

-- Export opportunities to CSV
COPY (
  SELECT * FROM leverage_opportunities
  WHERE created_at > NOW() - INTERVAL '48 hours'
) TO '/tmp/opportunities.csv' CSV HEADER;

Feed this to Gemini 2.0:

"Which of these opportunities were actually executable during high-volatility periods?"

---

⚠️ Disclaimer

This software is for educational and research purposes only.

• Trading with leverage involves significant risk of total capital loss
• All trades are simulated by default - no real execution
• Always test thoroughly before considering live trading
• Past performance does not guarantee future results
• Consult a financial advisor before making investment decisions

---

📝 License

MIT License - See LICENSE file for details

---

🤝 Contributing

Contributions welcome! Please open an issue or submit a pull request.

---