How to Trade Weather on Polymarket: Models, Cities and Resolution
A step-by-step weather trading workflow for Polymarket: pick the right city market, compare forecast models, price the threshold, size the position and avoid resolution and liquidity traps.

Weather markets look like the simplest category on Polymarket. The question is usually a temperature threshold in a named city on a specific date. The resolution source is public. The forecast models are free. That transparency is exactly why the category is competitive: everyone can see the same data, so the edge comes from interpretation and execution, not access.
This is the practical workflow. For the broader market mechanics, read the existing weather markets guide.
Step 1: Read the exact question
A weather market is not "will it be hot?" It is a precise question such as the highest temperature at a specific station on a specific date being above, below or between thresholds.
Check:
- the city and the station;
- the date and time window;
- the metric: high, low, average or precipitation;
- the unit: Fahrenheit or Celsius;
- the rounding rule;
- the resolution source;
- what happens if the source is unavailable.
The resolution documentation explains the general process. A one-degree difference or a different station can flip the outcome.
Step 2: Pick liquid city markets
Major cities with active weather markets usually have tighter spreads and more reliable data. Niche cities can offer mispricing but are harder to exit.
Check the order book:
- best bid and ask;
- depth within one or two cents;
- spread as a percentage of the price;
- whether the book is one-sided.
A market with a 10-cent spread is not a trading opportunity for a small edge. It is a liquidity tax.
Step 3: Build your own probability
Do not copy the market price and call it a forecast. Start with the models:
- the National Weather Service for U.S. stations;
- the ECMWF for medium-range guidance;
- local meteorological services for city-specific microclimates.
Compare:
- the deterministic forecast;
- ensemble spread;
- the forecast horizon;
- the model's recent bias for that station;
- the difference between the station and the city center.
A simple approach is to convert the ensemble into a probability of exceeding the threshold. If 70% of ensemble members are above 75°F, the market should be near 70 cents before risk premium and fees. If it trades at 55, you may have an edge; if it trades at 80, you may be late.
Step 4: Account for forecast uncertainty
The closer the threshold to the forecast, the more the outcome depends on model error. A market at 50 cents is often a coin flip around the threshold, not a mispricing.
Increase your edge estimate only when:
- the ensemble is tightly clustered;
- the threshold is far from the forecast;
- the station has a stable bias;
- the forecast horizon is short;
- the market price is far from your probability.
Step 5: Check resolution and manipulation risk
Weather markets have two special risks:
- Resolution ambiguity. The source may report a different value than the model or the city dashboard.
- Manipulation in thin markets. A small market can be moved by a large order, and suspicious activity around weather stations has been reported.
Trade liquid markets, read the rules, and avoid markets where the resolution source is unclear. If the market is thin, size for the exit, not just the entry.
Step 6: Size the position
Use a fraction of your bankroll based on the edge and the uncertainty. A practical rule:
- risk 1–3% of the bankroll per weather market;
- cap exposure per city and per date;
- avoid correlated positions across nearby thresholds;
- keep a cash buffer for new forecasts.
Weather markets are binary. A 70% probability still loses 30% of the time. Size so that a cluster of misses does not end the experiment.
Step 7: Execute with limit orders
Use limit orders where possible. A market order can sweep a thin book and turn a good forecast into a bad trade. If the price moves away, skip the trade.
Record:
- your probability;
- the market price;
- your fill;
- the model run time;
- the resolution value.
The difference between your probability and the market price is your edge estimate. The difference between your expected fill and actual fill is execution.
A weather trading checklist
- Exact station and date confirmed.
- Resolution source read.
- Ensemble probability estimated.
- Market price compared with probability.
- Spread and depth checked.
- Position size within risk budget.
- Correlated city/date exposure checked.
- Exit or hold-to-resolution plan written.
- Forecast update schedule set.
Common mistakes
- Trading the city forecast instead of the station.
- Ignoring rounding and time windows.
- Treating a 50-cent market as a strong signal.
- Using one model without ensemble spread.
- Sizing a thin market like a liquid one.
- Holding through a resolution source change.
- Adding positions after a forecast update without rechecking the price.
Bottom line
Weather trading is a forecasting discipline with a resolution wrapper. Read the exact question, build a probability from ensembles, compare it with the market, check liquidity and size for binary risk. The models are public; the edge is in doing the work consistently.
This article is research and education, not financial advice.
Sources
- 1.Resolution— Polymarket Documentation
- 2.National Weather Service— National Weather Service
- 3.ECMWF Forecasts— European Centre for Medium-Range Weather Forecasts
- 4.Prices and Orderbook— Polymarket Documentation