polymarket-quotebot
github.com/kamilch1k/polymarket-quotebot
Paper market-making bot that measures Polymarket LP rewards: reward share vs real book, fills vs real tape, and adverse-selection markouts with no orders placed and no API key required.
Market-making simulator that measures theoretical LP reward share, fill rates against real order book, and adverse selection costs using historical CLOB data without executing trades
This bot tackles a specific, underexplored problem on Polymarket: how profitable is being a liquidity provider under the CLOB reward system? Rather than trading, it simulates quote placement to calculate theoretical reward share using the actual order book. The code calculates three key metrics: (1) what fraction of total rewards your hypothetical quotes would earn based on spread and depth, (2) how often your quotes would have been filled vs the real tape, and (3) adverse selection markouts—how much the market moves against you after a theoretical fill. The architecture is clean: a data fetcher subscribes to CLOB WebSocket streams, a simulator runs quote logic at configurable refresh rates, and an analyzer produces summary statistics. The README explicitly states 'Phase 0 = no orders, no key,' meaning zero financial risk. The 33KB codebase is focused and well-commented. No malware signals detected. Created July 13, 2026, this is extremely fresh. It's ideal for anyone considering market-making on Polymarket who wants data-driven evidence before committing capital. The simulation approach could be extended to test custom quoting strategies. Recommend cloning for the analytical framework.