San Francisco weather data for Polymarket traders.

Polymarket weather markets resolve on a single number: the daily high temperature at a specific weather station, on a specific day, in San Francisco (KSFO). The market settles YES or NO depending on whether that number crosses a threshold. That sounds simple — but the share price between now and resolution is anything but stable.

Casual bettors price the market on intuition ("feels warm today"). Market makers price it on stale forecasts. Algorithmic flow is thin, especially during local night-time hours when America/Los_Angelesis asleep. The result: when the daily-max is already pushing the threshold by mid-afternoon, the YES share often hasn't fully repriced. That residual gap — between the share price and what the actual reading says — is the mispricing.

We compute that gap continuously for San Francisco. The current temperature is pulled from the KSFO station every five minutes. The daily-max-so-far is the running high since local midnight in America/Los_Angeles. The forecast for the rest of the day comes from a five-model ensemble — ECMWF, GFS, ICON, JMA, UKMO — with a per-city bias correction we fit on years of San Franciscoresiduals. From that we derive a probability that the close clears any given threshold. If Polymarket's implied probability disagrees, we tell you by how much.

We do not promise these markets are easy money. Resolution risk is real. Late-day heat waves can flip a market in the last hour. Forecasts miss. But across hundreds of weather markets we watch, the signal-to-noise on a measured probability gap is substantially better than buying the YES on a hunch.