Polymarket AI Trading Bot

This is a complete, production-ready AI trading agent designed to identify and exploit edge on Polymarket. It combines real-time market data with deep research and a systematic superforecasting engine to provide calibrated probability estimates and risk-managed trade recommendations.

🚀 2026 Upgrades: Autonomous Execution & Niche Alpha

The bot has been upgraded for the 2026 Polymarket landscape, focusing on high-signal niche markets and automated execution with a hardcoded Risk Shield.

Key New Features

Autonomous Execution: The place_order tool allows the bot to sign and post trades directly to the CLOB.
Risk Shield: Hardcoded $50 USDC per-trade cap and 2% slippage protection enforced at the tool level.
Niche Pivot: Targeted scanning of low-liquidity markets ($5k–$50k OI) where AI has a comparative advantage over speed-focused HFT bots.
Self-Ensemble Reasoning: Mandatory 3-persona reasoning (Bayesian, Sentiment, Domain) for more robust probability estimates.
Zen 2.0 TUI: A revamped, minimalist dashboard with high-density market grids and visual probability bars.
Performance Audit: High-precision telemetry tracking trade latency and automated Brier score calibration analysis.

Prerequisites

Node.js >= 18: Required for MCP servers and script utilities.
Python >= 3.9: Required for the Zen TUI dashboard and Audit script.
Gemini CLI: The interactive agent runner (Installation Guide).
Google AI Studio API Key: Required for Gemini model access (Get a key).
Optional: newsdata.io API key for live news feed integration.

Note: Run pip install -r requirements.txt to install Python dependencies.

Quick Start

Initialize the project:
```
bash scripts/setup.sh
```
Configure your environment: Open the newly created .env file and fill in your keys. For trading, you need POLYMARKET_PRIVATE_KEY (64-char hex or 44-char base64), POLYMARKET_API_KEY, POLYMARKET_API_SECRET, and POLYMARKET_API_PASSPHRASE.
Load the environment:
```
export $(grep -v '^#' .env | xargs)
```
Start the Dashboard:
```
python3 scripts/dashboard.py
```

Running Scripts

Niche Market Scan (Recommended): Focus on the long-tail niche edge.
```
bash scripts/scan_niche.sh
```

Standard Market Scan: Top 20 high-volume markets.

bash scripts/scan.sh --limit 10 --min-edge 0.05

Logical Inconsistency Check: Find mathematical price gaps in related markets.
```
node scripts/find_inconsistencies.js
```
Performance Audit: Analyze trade latency and prediction accuracy.
```
python3 scripts/audit_accuracy.py
```
Live Dashboard (Zen 2.0): Monitor bot status, market feed, and trades in real-time. Supports hotkeys: n (Niche Scan), s (Standard Scan), r (Refresh).
```
python3 scripts/dashboard.py
```

Project Structure

GEMINI.md: Foundational project context and operating mandates for the AI agent.
mcp/: Model Context Protocol servers for Polymarket and News APIs.
scripts/: Utilities for scanning, logic checking, auditing, and the Zen 2.0 TUI.
output/: Directory for scan results, predictions.jsonl, trades.jsonl, and audit_results.json.

Calibration & Accuracy

The bot tracks its own performance using the Brier Score $(p - o)^2$ and ensemble-level breakdowns.

Resolution: Running node scripts/resolve_predictions.js updates scores.
Audit: python3 scripts/audit_accuracy.py generates a full report on latency and calibration.

Risk Disclaimer

Prediction markets involve significant financial risk. This bot includes a hardcoded $50 Risk Shield, but you are responsible for any funds used. Always use a dedicated test wallet. This software is provided under the MIT License.

License

MIT

Polymarket Bot

Install / Use

README