Research
Data-driven analysis on prediction markets and quantitative forecasting.
Why 68% Accuracy Lost Money: Payoff Asymmetry in Prediction Markets
Our weather model was right 68% of the time and still lost money. The culprit: payoff asymmetry. When you pay 90¢ to win 10¢, even great accuracy can't save you. Here's how we found the one price band where the economics actually work.
How Accurate Are NOAA Weather Forecasts? We Tracked Every Degree.
We tracked NOAA high temperature forecasts against actual observations across Chicago, Denver, and Miami for a week. Average error: 1.3°F — but the city-by-city breakdown tells a more nuanced story about warm bias, flat terrain, and forecast revisions.
Phantom Liquidity: Why Our 0% Win Rate Was a Lie
Our weather trading bot showed 0% accuracy on "high edge" contracts. The data was lying — we were backtesting against phantom liquidity. Here's how fake order books distort prediction market strategy.
We Built a Weather Trading Bot. Here's What Day 1 Taught Us.
Our team built an automated weather trading system for Kalshi prediction markets. On Day 1, our "biggest edge" turned out to be wrong — but the real lesson was more interesting than the mistake.
New research published weekly.