I want to tell you about the dumbest edge we've ever found. It's weather markets.

Polymarket runs dozens of weather markets at any given time. "Will it be above 75°F in Phoenix on March 15?" or "Will NYC get more than 2 inches of rain this week?" Simple yes/no questions about the weather.

And they are consistently, laughably mispriced.

1The setup: NOAA vs. Polymarket

The National Oceanic and Atmospheric Administration (NOAA) publishes hourly forecasts for every major US city. These forecasts are very good. Within 24 hours, accurate to 2-3°F. Within 48 hours, usually within 4-5°F. This isn't some secret. It's public data anyone can pull from their API.

Real Example: Chicago Temperature Market

Market: "Will the high temperature in Chicago be above 50°F on Saturday?"

NOAA Forecast

58°F

~95% probability above 50°F

Polymarket Price

$0.65

Implies only 65% probability

That's a 30-cent edge on a dollar payout

2Why is the market so bad at weather?

A few reasons, and they're all structural:

Nobody cares

Weather markets have tiny volume, often $5K to $20K total. Serious traders focus on politics, crypto, and sports. Weather is a backwater.

People overweight uncertainty

Humans think "weather is unpredictable" and assign 60-70% odds, anchoring on general uncertainty rather than the specific, well-calibrated forecast.

No feedback loop

In political markets, pundits constantly discuss odds. Nobody publishes a "weather market newsletter." The market just sits there, mispriced, until resolution.

Capital lockup deters traders

Each trade earns a small profit ($0.30 on a $0.70 investment). To make real money, you'd need dozens of simultaneous positions. Most don't bother.

3Our approach: the pipeline

We built a system that checks NOAA forecasts every 5 minutes and compares them against live Polymarket prices for every active weather market. Here's the step-by-step:

Pull NOAA forecast

Get the official forecast for the relevant city and date.

Pull Open-Meteo as second opinion

Independent verification from a second forecasting source.

Agreement check

If the two models disagree by more than 5°F, skip, since the forecast isn't reliable enough.

Calculate probability

Use a normal distribution centered on the forecast value, with standard deviation calibrated to NOAA's historical accuracy.

Compare to Polymarket price

Calculate the edge between our probability estimate and the current market price.

Execute if edge ≥ 15%

Only trade the heavily underpriced side where the forecast is basically certain and the price doesn't reflect it.

Why the 15% minimum matters

We're not trying to be right 55% of the time for a slim edge. We only trade cases where the market is obviously wrong, where the forecast is basically certain and the price doesn't reflect it.

4The numbers

$200Max position size

$50Daily loss limit

15-20%Average edge

0Correlation to crypto

That might sound conservative, but weather is weather, and once in a while a freak cold front or heat wave will blow up a position. The $50 daily cap means one bad day doesn't wreck the month.

The edge on individual trades has been around 15-20% on average. Not every trade wins. We've been burned by sudden weather shifts that even NOAA missed. But the win rate has been high enough to make it consistently profitable.

The best part is that weather markets are almost perfectly uncorrelated with everything else. "Will it be above 80°F in Miami on Thursday?" doesn't care about the S&P 500.

Why uncorrelated alpha matters

When political markets crash because of an unexpected debate moment, or crypto markets dump on a whale liquidation, weather markets just keep ticking along. This makes weather a powerful diversifier in any prediction market portfolio.

5Will this edge last?

Probably not forever. If enough people start automating weather market trading, the prices will tighten. That's how markets are supposed to work: informed traders correct mispricings, and the edge evaporates.

But right now? Nobody's doing this at scale. Weather markets are too small and too boring for most traders to bother with. And as long as Polymarket keeps creating new weather markets faster than the few automated traders can cover them, the edge will keep appearing.

Current market state

New weather markets created

Faster than traders can cover them

We're enjoying it while it lasts.