- Kalshi’s January snowstorm contract topped $6 million in trading volume — one of the largest weather contracts ever on the platform — as AI-powered weather startups like Windborne Systems use prediction markets to test and improve their forecasting models
- An Interactive Brokers analysis found prediction market weather forecasts outperformed U.S. National Weather Service predictions — attributed to the financial incentive creating a “dual effect” of attracting accurate forecasters while penalizing inaccurate ones
- Scientists are building bespoke “sponsor-funded” prediction markets where reinsurers and climate research institutions supply the betting capital, using the aggregated forecasts to discover insurance-relevant climate risk beyond what traditional models can capture
- Critics warn of serious risks: data manipulation (a Ukraine war map was altered to resolve a Polymarket bet incorrectly), perverse incentives (reporters allegedly pressured to change stories), and the gamification of climate risk just as extreme weather makes some regions uninsurable
What Happened?
Prediction markets for weather events are growing from a niche curiosity into a serious financial and scientific tool. Kalshi’s contract on January’s New York megastorm reached $6 million in trading volume, and AI weather startups — including Windborne Systems and Swiss-based Jua — are now actively trading weather markets to test their forecasting models and generate returns. An analysis by Patrick Brown at Interactive Brokers found that prediction market weather forecasts outperformed official National Weather Service predictions, crediting the financial incentive structure for attracting more accurate forecasters. Meanwhile, researchers backed by French reinsurer SCOR are running sponsor-funded prediction markets specifically designed to aggregate expert climate knowledge for insurance pricing purposes — a model closer to the original academic vision for prediction markets than the retail gambling boom on Kalshi and Polymarket.
Why It Matters?
Weather derivatives have existed since the 1990s, but they’ve always suffered from low liquidity and narrow participation. Prediction markets offer a potential solution by opening the pool to casual participants, AI systems, and domain experts simultaneously — potentially generating better-priced risk signals for insurers and investors who currently rely on proprietary models. The climate context makes this increasingly urgent: the last 11 years were the hottest on record, extreme weather is making some regions uninsurable, and traditional parametric insurance products are expensive and slow-moving. But the gamification of weather data carries real downside risks. A Ukraine war map was manipulated long enough to incorrectly resolve a Polymarket contract, and Israeli reporters say Polymarket users pressured them to change coverage to affect bet outcomes — warning signs that financial incentives can corrupt the information these markets are supposed to aggregate.
What’s Next?
The most intellectually serious version of weather prediction markets — scientist-run, sponsor-funded pools designed to extract expert judgment on hurricane frequency and El Nino timing — is advancing through institutions like CRUCIAL at Lancaster University, backed by SCOR Foundation. That model deliberately separates itself from retail gambling by having sponsors absorb the financial losses in exchange for better forecasts. The retail end of the market is meanwhile scaling rapidly: WeatherBook, founded by a former CME product strategist, is building a dedicated weather prediction market platform. As climate volatility increases and insurance markets crack under the strain, the gap between what traditional models can price and what expert prediction markets might reveal is likely to grow — making this one of the more consequential experiments at the intersection of finance and climate science.
Source: Bloomberg
