Polymarket AI Arbitrage Bot
Budget: $250 – $750 USD
I want a Python-based bot that can watch Polymarket in real time, spot profitable price gaps, and place trades automatically the moment an edge appears. My goal is hands-free arbitrage and “copy” execution that mirrors the best available positions without delay.
Core requirements
• Automated trading only—once the parameters are set, the system must handle order sizing, wallet management, and execution from start to finish.
• Machine-learning decision layer that learns from past outcomes to refine which markets to target, when to enter, and when to exit.
• Direct integration with the official Polymarket API or websocket feed for live odds, order-book depth, and settlement updates.
• Robust error handling: failed transactions need graceful retries; connection drops should automatically reconnect and sync state.
• Simple configuration file or dashboard where I can adjust risk limits, maximum stake per market, and profit-taking thresholds.
• Clean, well-commented Python code plus a README that lets me install dependencies, set environment variables, and run the bot on Linux or a cloud VM.
Acceptance criteria
1. The bot maintains a persistent connection for 72 hours without manual intervention while executing at least ten trades.
2. Net return is positive on a representative sample account, or a detailed log explains missed opportunities and decision rationale.
3. All trades are recorded to disk (CSV or SQLite) with timestamp, market, position, stake, outcome, and realized PnL.
4. I can reproduce the environment using only the README instructions.
If you have existing libraries for on-chain signing, gas optimization on Polygon, or reinforced learning frameworks (TensorFlow, PyTorch, etc.), feel free to leverage them as long as the final deliverable meets the above points.
Core requirements
• Automated trading only—once the parameters are set, the system must handle order sizing, wallet management, and execution from start to finish.
• Machine-learning decision layer that learns from past outcomes to refine which markets to target, when to enter, and when to exit.
• Direct integration with the official Polymarket API or websocket feed for live odds, order-book depth, and settlement updates.
• Robust error handling: failed transactions need graceful retries; connection drops should automatically reconnect and sync state.
• Simple configuration file or dashboard where I can adjust risk limits, maximum stake per market, and profit-taking thresholds.
• Clean, well-commented Python code plus a README that lets me install dependencies, set environment variables, and run the bot on Linux or a cloud VM.
Acceptance criteria
1. The bot maintains a persistent connection for 72 hours without manual intervention while executing at least ten trades.
2. Net return is positive on a representative sample account, or a detailed log explains missed opportunities and decision rationale.
3. All trades are recorded to disk (CSV or SQLite) with timestamp, market, position, stake, outcome, and realized PnL.
4. I can reproduce the environment using only the README instructions.
If you have existing libraries for on-chain signing, gas optimization on Polygon, or reinforced learning frameworks (TensorFlow, PyTorch, etc.), feel free to leverage them as long as the final deliverable meets the above points.