Weather Market NO-Side Overpricing Exploitation
Edge sourced from systematic overpricing of YES outcomes in Polymarket temperature markets, identified by comparing market prices to physically-grounded meteorological probability models.
This strategy targets Polymarket markets that ask whether a specific location's temperature will exceed a threshold on a given date. The bot computes a Gaussian probability distribution over the expected temperature using live METAR (aviation weather observation) data and Open-Meteo forecasts, then compares the model's NO probability to the market's implied NO price. When the market overprices the YES outcome (i.e., underprices NO), the bot buys the NO side.
The Gaussian model is physically motivated: temperature distributions around forecasts are approximately normal, and the model can precisely estimate the probability of exceeding any threshold given the forecast mean and historical variance. METAR data provides high-frequency ground-truth observations to update the forecast in near real-time as the resolution date approaches.
Entry conditions require the model's estimated NO probability to exceed the market's NO price by a configurable margin. The strategy operates in paper-trading mode, suggesting the author is still in the validation phase. The bet is directional (always NO) based on the thesis that retail participants systematically over-bet extreme or exciting temperature events.
Key risks include model error from non-Gaussian temperature distributions during extreme weather events, forecast errors from Open-Meteo in unusual meteorological conditions, market illiquidity limiting position sizes, and the possibility that Polymarket temperature markets attract sophisticated meteorological traders who keep prices well-calibrated. The always-NO bias is also a risk if the thesis about retail overpricing is wrong or market-specific.