.svg)
.svg)
.svg)
How to Trade Polymarket Weather Markets: A Complete Guide
Most guides to prediction market trading skip the part that actually matters: not how to place a trade, but how to decide whether a trade has positive expected value in the first place. This guide covers both — the mechanics of getting set up on Polymarket and the framework for making weather trades that aren't just expensive guesses.
Before You Start: What You're Actually Doing
Trading Polymarket daily temperature markets is not betting on the weather. It is expressing a view that the current market price for a temperature bucket is wrong — specifically, that the true probability of that outcome is higher (or lower) than what the market currently reflects.
If the NYC 68–69°F bucket is priced at $0.30 and you believe the genuine probability is 45%, that's a +EV trade. If you have no independent view — you're just going with what feels right — you are providing liquidity to people who do have independent views. That rarely goes well.
The foundation of profitable weather trading is a forecast. Everything else is mechanics.
Step 1: Check Your Geographic Eligibility
Polymarket restricts access for traders in approximately 33 countries as of 2026, including the United States, United Kingdom, France, Germany, and most other Western European nations.
The geographic restriction is enforced by IP address detection, device fingerprinting, and additional signals. The Terms of Service explicitly prohibit VPN circumvention with the risk of account suspension and frozen funds.
Step 2: Set Up Your Wallet and Funding
Polymarket runs on the Polygon PoS blockchain and settles all positions in USDC.e (bridged USDC). You do not need to hold ETH or any other asset to trade.
Option A: Email login (simplest) — Sign up at polymarket.com with an email address. Polymarket creates a Magic.link-powered proxy wallet for you automatically.
Option B: MetaMask or hardware wallet (more control) — Connect any EIP-1193 compatible wallet. To deposit funds, send USDC.e to your Polygon address directly — either via a Polygon-compatible exchange withdrawal (Kraken, Coinbase, etc.) or by bridging USDC from Ethereum using the official Polygon bridge.
Gas fees: Polygon transactions cost fractions of a cent in POL. Keep a small amount of POL in your wallet — roughly $2–$5 worth covers weeks of gas at current prices.
Minimum position size: Trades below $5 are economically meaningless given spread costs. Most serious manual traders start with at least $100 per meaningful position.
Step 3: Find a Weather Market
Navigate to polymarket.com/climate-science/weatheror search for “temperature” in the Markets page. Each market shows the question, resolution source, all outcome buckets with current YES prices, total volume, and end time.
Before placing anything, filter for:
- Liquidity — Markets with less than $1,000 in volume often have spreads of 5–10¢ per share. NYC, London, Shanghai, and Tokyo have the highest volume.
- Time to expiry — Markets more than 72–96 hours out require meaningful forecast skill at longer lead times, which most retail traders do not have a reliable edge in.
- Identifiable resolution station — Read the market description to confirm you know exactly which airport station resolves this market.
Step 4: Read the Forecast — This Is the Core Skill
Before you look at the market price, look at the forecast. This is non-negotiable. You need a temperature forecast specific to the resolution station, not the city center. The resolution station is an airport. Most consumer weather apps report city-center readings. These can differ by 3–8°F.
Free tools that give airport-station data:
- Weather Underground (wunderground.com) — Search the ICAO station code (e.g., KLGA for LaGuardia, LFPB for Paris-Le Bourget). Wunderground is the data source Polymarket uses for resolution.
- aviationweather.gov — Official FAA tool for METAR data and TAF (Terminal Aerodrome Forecast) for the next 24 hours.
- Open-Meteo.com — Free API providing GFS, ECMWF, and other ensemble output at specific coordinates. You can search by airport name or lat/lon.
At 24 hours out, a well-calibrated short-range forecast (HRRR, NBM, NAM) has a typical error of 2–3°F. At 48 hours, 3–4°F. At 72 hours, 4–6°F. These error ranges are your uncertainty budget — they tell you how spread out the probability mass should be across the buckets.
Step 5: Translate the Forecast Into a Probability
This is the step most retail traders skip, and it's why they lose. A deterministic forecast (e.g., “high of 68°F”) does not tell you which bucket to bet. It tells you the most likely single value. But the daily high is not a point estimate — it's a distribution.
Simple Method: Normal Distribution Approximation
Assume the forecast is unbiased and the errors are roughly normally distributed. Use the forecast error appropriate for the lead time as your standard deviation (σ). For example: 36 hours out, σ ≈ 2.5°F. Forecast high: 67°F.
A Normal distribution centered at 67°F with σ = 2.5°F assigns these approximate probabilities:
| Bucket | Model Probability |
|---|---|
| Below 62°F | ~2% |
| 62–63°F | ~5% |
| 64–65°F | ~14% |
| 66–67°F | ~25% |
| 68–69°F | ~25% |
| 70–71°F | ~14% |
| Above 72°F | ~8% |
This is your model-implied distribution. Compare each bucket's model probability to its current market price. Any bucket where the market price is materially lower than your model probability is a candidate long.
Better Method: Use Ensemble Forecasts
Instead of assuming a standard deviation, use the actual spread from an ensemble forecast. Open-Meteo's Ensemble API gives you 51 ECMWF members or 31 GFS members, each with a different forecast for the daily high. Count how many members land in each bucket. That fraction is your empirical probability estimate — no distributional assumption required.
Step 6: Check the Edge Before You Trade
The edge on a trade is the difference between your model probability and the market price.
Edge (%) = (Your probability − Market price) × 100
An 8% edge means you believe the true probability is 8 percentage points higher than what you're paying. That's a meaningful edge for a daily temperature market.
Why 8%?
- Polymarket's spreads eat 1–3% of EV depending on the market.
- Forecast uncertainty means your own model has error — you need buffer.
- You want to be making bets where the math clearly favors you, not marginal ones.
If your edge is less than 5%, pass.
Step 7: Size the Bet with Kelly Criterion
The mathematically correct position size for a binary bet with edge is given by the Kelly Criterion:
f = (q − p) / (1 − p)
Where q = your probability estimate, p = the market price, and f = fraction of bankroll to bet.
Example:Market price $0.30, your probability 0.45. f = (0.45 − 0.30) / (1 − 0.30) = 0.15 / 0.70 ≈ 21% of bankroll. That's full Kelly — most practitioners use ¼ to ½ Kelly in practice, giving 5–10% of bankroll.
Absolute caps matter more: Even if Kelly says 15%, cap individual positions at a fixed dollar amount — say $50–$100 — until you have a statistically meaningful track record that validates your forecast model.
Step 8: Place the Trade
Click into the temperature market. Click the bucket (outcome) you want to buy. Select YES or NO.
Limit orders are usually preferable for manual weather trades — you specify your price and wait for a seller to accept it, avoiding the spread. Market orders buy at the best available ask price: fast, but you cross the spread and pay full ask.
Common Mistakes That Cost Money
- Using a city-center forecast. The market resolves at an airport. A weather app set to “London” shows Westminster. The resolution station is London City or Heathrow.
- Trading without an edge calculation. Gut feel that “it'll be warm tomorrow” is not a trading thesis. Translate your forecast into a probability first.
- Full Kelly sizing. The Kelly Criterion assumes your probability estimate is perfectly calibrated. It's not. Use fractional Kelly and hard position caps.
- Trading illiquid markets. A 7¢ spread on a $0.30 contract is a 23% round-trip cost. Stick to liquid markets until you have demonstrated edge.
- Chasing the final-hour spike. Entering at $0.85 on a bucket with 90% probability leaves almost no upside and meaningful downside if the reading comes in at a boundary.
What Consistent Profit Actually Looks Like
Weather trading at scale is a volume game. Individual bets have significant variance — even a 65% probability trade loses 35% of the time. The edge compounds over hundreds or thousands of bets, not individual outcomes.
The traders on Polymarket's weather leaderboard with five- and six-figure profits traded thousands of times across dozens of cities and multiple lead times. They built a statistical advantage through better probability estimates, consistent sizing, and disciplined execution, then repeated it continuously.