An automated system that finds and exploits pricing inefficiencies in prediction markets.
Prediction markets (Polymarket, Kalshi) let you bet on real-world events - "Will BTC hit $100k?", "Will it rain in Miami tomorrow?". These markets often lag behind reality:
- Crypto prices move on Binance seconds before prediction market odds adjust
- Weather forecasts update before betting odds reflect new data
- The same event priced differently across platforms (45¢ vs 52¢)
This lag creates arbitrage opportunities.
Eight autonomous agents working together:
| Agent | Job |
|---|---|
| Venue Watchers | Stream prices from Polymarket, Kalshi |
| Oracles | Stream real-world data (Binance crypto, FRED economic data, NWS weather) |
| Opportunity Scanner | Detect when markets lag reality |
| Strategy Agents | Compete for capital using different approaches |
| Risk Guardian | Veto power - enforces position limits, drawdown stops |
| Executor | Places trades, handles fills |
- Initial capital = risk money (start with $500)
- Profits are protected via ratcheting stop-loss
- Grow to $800? New floor is $640 (80% of peak)
- Full automation with human-controlled kill switch
- BTC jumps 2% on Binance
- Polymarket "BTC up in next 15 min" still priced at 45¢
- Bot buys YES at 45¢
- Market adjusts to 65¢
- Bot sells or holds to resolution
- Language: Python 3.12
- Message Bus: Redis Streams
- Database: PostgreSQL
- Dashboard: Streamlit
- Deployment: Railway
| Sprint | Status | Deliverable |
|---|---|---|
| 1 | Complete | Foundation - agents communicate via message bus |
| 2 | In Progress | Live data from Polymarket, Kalshi + oracles (Binance, FRED, NWS) |
| 3 | Planned | Opportunity detection |
| 4 | Planned | Risk Guardian + paper trading |
| 5 | Planned | Strategy tournament system |
| 6 | Planned | Dashboard |
| 7 | Planned | Go live |
# Start infrastructure
docker-compose up -d
# Install dependencies
pip install -e ".[dev]"
# Copy and configure environment
cp .env.example .env
# Edit .env with your API keys (Polymarket, Kalshi, FRED, etc.)
# Run tests
pytest tests/ -v
# Run linter
ruff check src/ tests/Note on US legal compliance: Kalshi is a CFTC-regulated exchange available to US residents. Polymarket has restrictions for US-based users. Ensure you comply with local regulations before enabling live trading on any venue.
- Design Document - Full architecture and rationale
- Implementation Plan - Sprint-by-sprint tasks
Private project.
