Kalshi Releases Comprehensive Study on Prediction Market Accuracy

Kalshi has unveiled what it describes as the largest analysis of prediction market accuracy to date, encompassing over 300,000 contracts with detailed transaction-level data. The study, conducted by researchers Bürgi, Deng, and Whelan, concludes that prediction market prices serve as genuinely informative estimators of real-world probabilities, although it warns that purchasing low-priced contracts can lead to substantial losses.

The analysis indicates that contract prices closely track actual outcomes, with a contract trading at 70 cents correlating to a 70% likelihood of the event occurring. Notably, the accuracy of these predictions improves significantly as the contract approaches its expiration date, with mean absolute pricing errors decreasing sharply on the final trading day. Earlier in a contract's lifespan, prices tend to be noisier and less reliable.

One key finding highlights the 'favorite-longshot bias,' where contracts priced at 10 cents or less, indicating low probability events, tend to have even lower success rates than expected. Conversely, contracts representing favored outcomes, while delivering smaller returns, tend to be slightly underpriced.

This study is part of Kalshi Research, a division established to analyze probability calibration and compare prediction markets against traditional forecasting methods. Kalshi’s inflation forecasts reportedly outperformed Wall Street predictions by 40%, underscoring the credibility of its research, especially given its regulation by the CFTC.

In a related development, CME Group CEO Terry Duffy recently clashed with CFTC Chairman Michael Selig at a meeting regarding the integrity of prediction markets. Duffy argued that some contracts could be manipulated, a claim Selig dismissed as 'fake news.' This exchange highlights ongoing tensions in the regulatory landscape surrounding prediction markets.