Polymarket weather bucket arbitrage bot
Exploits systematic edge in Polymarket daily weather bucket markets across three temperature categories using live METAR observations and Open-Meteo forecasts.
This bot runs an automated paper-trading strategy on Polymarket's weather bucket markets, betting the NO side when markets overprice a temperature outcome. It uses live METAR observations and Open-Meteo forecasts with a Gaussian probability model to estimate true probabilities and identify mispriced markets.
Entry: When the implied probability from the market price exceeds the model's estimated probability by a calibrated threshold, the bot places a NO bet (selling the YES share). Exit: The position is held until market resolution or until the mispricing reverts, with dynamic position sizing based on edge size.
Verified results: The repository shows a logically sound paper-trading framework with 4 stars. No real-money PnL is provided but the approach is analytically rigorous.
Implementation notes: Written in Python, uses public weather APIs and Polymarket CLOB API. Risks include model error, market liquidity issues, and resolution risk from weather data discrepancies.