Prediction market traders consistently outperform professionals at predicting inflation, especially when readings deviate significantly from expectations, according to a study by prediction markets Kalshi.
According to a report shared with CoinDesk, Kalshi compared inflation forecasts on its platform to Wall Street consensus forecasts and found that over a 25-month period, market-based traders were more accurate than traditional economists and analysts, especially during periods of economic volatility.
The research found that market-based estimates of year-on-year changes in the consumer price index (CPI) showed an average error between February 2023 and mid-2025 that was 40% lower than consensus forecasts. The differences are even more pronounced when numbers deviate significantly from expectations. Under these circumstances, Kalshi’s forecast beat the consensus by 67%.
The study, “Crisis Alpha: When do prediction markets surpass expert consensus?” also examined the relationship between the size of forecast disagreements and the likelihood of surprises.
When Kalshi’s CPI forecast deviated from the consensus by more than 0.1 percentage point in the week before release, the chance of a material deviation in the actual CPI reading rose to about 80%, compared with a baseline of 40%.
Unlike traditional forecasts, which often reflect a shared set of models and assumptions, prediction markets such as Kalshi and Polymarket aggregate forecasts from individual traders who have a financial incentive to accurately predict outcomes.
Kalshi’s user base has grown recently with the integration of prediction markets into major cryptocurrency wallet Phantom. The company raised $1 billion earlier this month at an $11 billion valuation, as bets on prediction markets continue to grow. In October, Polymarket was said to be in talks to raise capital at a valuation of up to $15 billion.
The report’s authors note that while the sample of large-scale shocks is relatively small, the data point to the potential role of market-based forecasts as part of broader risk and policy planning tools.
“Although the sample size of shocks is small (as it should be in a largely unexpected world), the pattern is clear—the market’s information aggregation advantage becomes most valuable when the forecasting environment becomes most challenging,” the study said.
Earlier this year, research by a data scientist showed that Polymarket was 90% accurate at predicting how an event would occur a month later, and 94% accurate at predicting hours before the actual event. Nonetheless, tacit bias, herd mentality, and low liquidity can lead to overestimation of event probabilities.
Why prediction markets outperform consensus during times of stress may depend on how they aggregate information. Research shows that traditional forecasts often rely on similar data and models across agencies, which can limit their ability to react when economic conditions change.
In contrast, prediction market platforms reflect the views of diverse traders, which draw on a range of inputs from specific industry trends to alternative data sets, creating what the study describes as a “wisdom of crowds” effect.
Incentives are also different. Institutional forecasters face reputational and organizational constraints that can hinder bold predictions. However, traders in prediction markets own money and are rewarded or punished purely based on performance.
The continuity of market pricing updated in real time also avoids the lag in consensus estimates, which are often set days before data is released.
“Instead of wholesale replacement of traditional forecasting methods, agency decision makers may consider market-based signals as complementary information sources of particular value in times of structural uncertainty,” the study states.
