Skip to content
This repository was archived by the owner on Apr 25, 2026. It is now read-only.

Latest commit

 

History

2 Commits

Folders and files

NameName
Last commit message
Last commit date
 
 
 
 

Repository files navigation

Weather Edge

AI-powered weather predictions for Kalshi prediction markets.

What is this?

Weather Edge uses a multi-model ensemble (31 GFS + 51 ECMWF members) with NWS bias correction to generate daily temperature predictions for Kalshi weather markets. When our model finds a statistically significant mispricing (8%+ edge), it generates a trading signal.

Stats

  • 82 weather models combined into probability distributions
  • 1.8°F Mean Absolute Error on 7-day historical validation
  • 14 US cities covered (LAX, MIA, CHI, AUS, DEN, BOS, DAL, NYC, PHX, SEA, SFO, ATL, HOU, MIN)
  • 3 daily scans (5 AM, 7 AM, 9 AM CT) — most edge found pre-10 AM

Track Record

Metric Value
Safety-regime win rate 88% (7W / 1L)
Full sample 50% (8W / 8L, includes pre-calibration)
First live trade DAL HIGH NO — WIN +$0.66 (Apr 10, 2026)
Model accuracy 1.8°F MAE, 96.9% at 90%+ confidence

How it works

  1. Ensemble forecasting: Combines GFS (31 members) + ECMWF (51 members) into temperature probability distributions
  2. NWS bias correction: Corrects systematic model errors city-by-city using NWS point forecasts (70% weight)
  3. Market comparison: Compares model probabilities against Kalshi contract prices
  4. Signal generation: Flags contracts where model probability diverges from market price by 8%+ (our edge threshold)

Newsletter

Get free daily signals before the market opens: stewyboy1990.github.io/weather-edge

Disclaimer

This is not financial advice. Past performance does not guarantee future results. Prediction market trading involves risk of loss. Trade responsibly.

About

AI Weather Predictions for Kalshi Prediction Markets

Topics

Resources

Stars

1 star

Watchers

0 watching

Forks

Releases

Packages

Contributors

Languages