icabrera0/Polymarket-paper-trader

Polymarket Paper Trading Bot is an autonomous paper-trading system for Polymarket prediction markets. It ingests live news (NewsAPI, GDELT, Telegram), uses an LLM panel (Anthropic Claude or local Ollama) to infer outcome probabilities, identifies mispriced markets via a synthesis of three agent opinions, and simulates trades under configurable risk rules—no real funds are used. Intended users: quant researchers, strategy developers, and traders who want to test LLM-driven signals and execution logic safely. Key features: - Multi-source real-time news ingestion with fuzzy deduplication. - 3-agent LLM panel (Quant, Domain, Adversary) + synthesis; computes model prob, confidence, Z-score, EV. - Market scanner for Polymarket Gamma/CLOB; filters by volume, spread, expiry. - Position sizing via quarter-Kelly with a 5% cap; parametric 95% VaR checks; slippage guard; stop-loss/TP rules; kill-switch. - Backtester (current and replay modes), Streamlit dashboard, daily Excel reports, Discord notifications. - SQLite persistence, APScheduler, full pytest suite. Requires Python 3.10+ and Anthropic key or local Ollama.

AI-related keywords found in this repo's description, topics, or README summary — grouped by category. Each badge links to the corresponding ranking detail page.