Kalshi says the largest analysis of prediction-market accuracy to date, covering more than 300,000 contracts with transaction-level data, found that market prices are informative estimators of real-world probabilities. Contracts trading at 70 cents corresponded to events occurring roughly 70% of the time, although pricing was less reliable earlier in a contract’s life and mean absolute errors fell sharply during the final day before resolution. The study by researchers Bürgi, Deng, and Whelan also identified a favorite-longshot bias: contracts priced at 10 cents or less won even less often than their prices implied, producing average buyer losses of more than 60%, while expensive contracts tied to heavily favored outcomes generated small average gains. Kalshi Research, established on December 22, 2025, said its mandate includes probability calibration, comparisons with traditional forecasting methods, and analysis of participant behavior. An early claim from the division said Kalshi’s inflation forecasts outperformed Wall Street consensus predictions by 40% across various conditions. Kalshi’s CFTC-regulated status gives the research greater institutional credibility than data from offshore or crypto-native platforms, while the dedicated research arm positions the company as both a trading venue and a data and analytics provider. The dataset may set a benchmark for future prediction-market research, which has often drawn on smaller samples from platforms including Intrade and the Iowa Electronic Markets.