The weather market on Polymarket was hacked by Chinese traders, turning it into a printing press.

$240 principal, $83k profit. His bot does only one thing: read weather data.
No hedge funds, no insider data sources, no climate PhDs. Just a coder, a script, a free weather API.
I spent a whole day reverse-engineering his script, and it left me hair-raising:
> No massive weather database
> No satellite cloud images, no radar interfaces
> Not even “rocket science” level climate models
His Polymarket homepage:

The core logic breaks down into three layers:
1. He doesn’t predict the weather; he arbitrages mispricings in the market
Polymarket asks: “Will the high temperature in New York be 52–54°F tomorrow?”
Most gamblers → guess based on intuition.
NOAA weather forecast → 85–90% accuracy.
The bot compares the two.
Market price 15 cents, forecast shows 45% probability.
→ Free money.
He’s not betting on the weather; he’s selling “overhyped uncertainty” to gamblers.
2. Multi-model voting, not trusting a single data source
He never relies on just one weather model.
The bot pulls four data sources simultaneously:
• GFS
• ECMWF
• UKMO
• NWS
Out of 31 model calculations, 28 say “above 70°F” → true probability 90%.
And at this point, the market price is only 60 cents?
→ Crushing advantage.
The bot only trades when the difference between the true probability and the market price exceeds 5–8%.
No advantage, no trade.
3. Dirty details: airport calibration
This detail can trap 99% of people trying to imitate him.
> The settlement for the New York market is based on LaGuardia Airport
> Dallas uses Love Field Airport
> Not the downtown temperature
The temperature difference between airports and city centers is often 3–8°F.
For narrow markets of 1–2°F, this deviation is the distance between heaven and hell.
He knows. Others don’t.
The rest is scaled-up rolling:
• Hundreds of micro-trades
• Over 60 cities
• Using Kelly formula to calculate position size
Each trade only earns a few dollars.
But compounded daily, over 10 months.
$240 → $83k.
Markets aren’t impossible to beat with science; they just react too slowly.