Kalshi Research studies found that prediction-market prices broadly reflected real-world probabilities, with a contract priced at 70 cents resolving yes about 70% of the time. Accuracy improved with time, market depth, participation and trading volume, while mean absolute errors fell sharply during the final day before resolution. The research found no empirical evidence that demographic bias, intentional manipulation or thin trading caused significant systemic distortion. It also identified a favorite-longshot bias: contracts priced at 10 cents or less won less often than implied and produced average buyer losses above 60%, while expensive contracts tied to heavily favored outcomes generated small average gains. Elections, politics and finance produced the strongest forecasts, and markets with about $10,000 in volume or as few as 20 traders still generated useful forecasts. Kalshi says its research division, established Dec. 22, 2025, covers calibration, comparisons with traditional forecasts and participant behavior; an early claim said its inflation forecasts outperformed Wall Street consensus predictions by 40% across various conditions.