Benchmark Llms On Prediction-market Decision Quality (brier, Calibration, Alpha)

MCP server and CLI that gives AI agents $10k paper money to trade real Polymarket order books, with level-by-level fills, slippage and fee modeling, strategy backtesting, P&L stats and a leaderboard. Companion benchmark harness scores LLMs on prediction markets.

MCP | Python · mcp-server · polymarket