Skip to content

Latest commit

 

History

History
261 lines (196 loc) · 6.55 KB

README.md

File metadata and controls

261 lines (196 loc) · 6.55 KB

Triangular Arbitrage Trading System

Automated cryptocurrency arbitrage detection and execution across CEX and DEX exchanges.

Features

  • CEX Arbitrage: Multi-exchange support (Binance, Kraken, KuCoin, Coinbase)
  • DEX Arbitrage: Uniswap V2/V3 cross-DEX opportunities on Ethereum/Base
  • Decision Engine: Explicit EXECUTE/SKIP logging with detailed reasoning
  • Web Dashboard: React + FastAPI real-time monitoring and control
  • Paper Trading: Risk-free testing with live market data
  • ML Optimization: GNN-based cycle scoring learns from execution history

Quick Start

Installation

git clone <repository-url>
cd Triangular-Arbitrage
pip install -r requirements.txt
cp .env.example .env

Running

Interactive Menu (Recommended)

python run_clean.py

CEX Arbitrage

# Paper mode (safe)
python run_clean.py cex --paper

# Live trading (requires API keys)
python run_clean.py cex --live

DEX Arbitrage

# Paper mode with live chain data
python run_clean.py dex --quiet

# Single scan test
python run_clean.py dex --quiet --once

# Live mode (requires wallet setup)
python run_clean.py dex --live --config configs/dex_mev.yaml

Web Dashboard

# Start server
python web_server.py

# Access at http://localhost:8000
# Build frontend: cd web_ui && npm run build

Configuration

Production Strategy Configuration

For live trading, use the production example configuration:

# Copy the production example
cp configs/strategies/strategy_production_example.yaml configs/strategies/my_strategy.yaml

# Edit your strategy
nano configs/strategies/my_strategy.yaml

Key Configuration Options:

# Price Mode - Use bid/ask for more accurate profit estimation
use_bid_ask: true  # Recommended for live trading (more conservative)
                   # Set to false to use last price (faster but less accurate)

# Profit Thresholds
min_profit_bps: 20          # Minimum net profit (0.20% = 20 basis points)
max_slippage_bps: 30        # Maximum allowed slippage per leg
max_leg_latency_ms: 3000    # Maximum execution time per leg

# Safety Margins
safety_margin_bps: 10       # Extra buffer for market fluctuations
spread_buffer_bps: 5        # Additional buffer for wide spreads

# Feature Flags - Disable experimental features
feature_flags:
  enable_dex_arbitrage: false    # Only enable if fully tested
  enable_mev_solver: false       # Only enable if ready
  enable_precomputed_cycles: false  # Use cycle files for speed

Environment Variables

Edit .env for basic settings:

# Trading
MAX_POSITION_SIZE=100
MIN_PROFIT_THRESHOLD=0.5
PAPER_USDT=1000

# Display
VERBOSITY=normal
TOPN=3

# Symbol filtering
SYMBOL_ALLOWLIST=BTC,ETH,SOL,AVAX,LINK
TRIANGLE_BASES=USD,USDT

For DEX, configure configs/dex_mev.yaml:

rpc_url: "https://mainnet.infura.io/v3/YOUR_KEY"
max_position_usd: 1000
threshold_net_pct: 0.1

Decision Engine

Every opportunity is evaluated and logged:

[16:30:15] Decision EXECUTE reasons=[] metrics: gross=0.80% net=0.40% breakeven=0.60% fees=0.30% slip=0.05% gas=0.05% size=$1000.00

[16:30:20] Decision SKIP reasons=[threshold: net 0.04% < 0.20%] metrics: gross=0.39% net=0.04% breakeven=0.55% fees=0.30% slip=0.05% gas=0.20% size=$100.00

Debug API:

# Check why trades aren't executing
curl http://localhost:8000/api/dex/decisions | jq '.decisions[:5]'

# Review rejection patterns
curl http://localhost:8000/api/dex/decisions | jq '.decisions[] | select(.action == "SKIP") | .reasons'

Architecture

┌─────────────────┐
│  React Frontend │  Web dashboard (port 3000/8000)
└────────┬────────┘
         │ WebSocket + REST
┌────────▼────────┐
│  FastAPI Server │  Real-time updates, decision tracking
└────────┬────────┘
         │
┌────────▼────────┐
│ Trading Engines │  CEX (trading_arbitrage.py)
│                 │  DEX (run_dex_scanner in web_server.py)
└─────────────────┘

Key Components:

  • decision_engine.py - Unified trade execution decisions
  • trading_arbitrage.py - CEX arbitrage engine
  • web_server.py - FastAPI backend with DEX scanner
  • web_ui/ - React dashboard

Testing

# Decision engine tests
pytest tests/test_decision_engine.py -v

# All tests
pytest -v

# With coverage
pytest --cov=triangular_arbitrage

Web Dashboard

Start Backend

python web_server.py

Build Frontend

cd web_ui
npm install
npm run build
cd ..

Access

Open http://localhost:8000

Features:

  • Real-time opportunity feed
  • Trade history and fills
  • Equity curve tracking
  • Decision trace debugging
  • System logs
  • Control panel (start/stop, configure)

API Endpoints:

  • GET /api/dex/status - Current status with last decision
  • GET /api/dex/opportunities - Opportunity feed
  • GET /api/dex/fills - Recent fills
  • GET /api/dex/decisions - Decision history (debugging)
  • POST /api/dex/control - Start/stop with config

Documentation

Safety

Built-in Protections:

  • Position limits
  • Profit thresholds
  • Depth validation
  • Slippage estimation
  • Decision engine gates
  • Paper mode testing

Always:

  • Test in paper mode first
  • Use small position sizes initially
  • Monitor decision logs
  • Check rejection reasons
  • Review /api/dex/decisions endpoint

Troubleshooting

No trades executing:

# Check decisions
curl http://localhost:8000/api/dex/decisions | jq -r '.decisions[] | "\(.action) \(.reasons | join(", "))"'

# Lower threshold if too many "threshold" rejects
curl -X POST http://localhost:8000/api/dex/control -H "Content-Type: application/json" -d '{"action": "start", "mode": "paper_live_chain", "config": {"min_profit_threshold_bps": 5}}'

Size issues:

  • Minimum: $10 per trade
  • Per-leg minimum: $5 (CEX)
  • Adjust size_usd in config

See WEB_DASHBOARD_README.md "Debugging Decisions" section for comprehensive troubleshooting

License

MIT License

Disclaimer

⚠️ For educational purposes only. Cryptocurrency trading carries substantial risk. Always test in paper mode first. Never trade with money you cannot afford to lose. Use at your own risk.