Prediction Markets Beat Wall Street on Inflation, Kalshi

New research by Kalshi on prediction markets suggests that those platforms can outpace traditional Wall Street estimates when it comes to forecasting US inflation.

Kalshi study finds prediction markets outperform Wall Street on inflation data

According to a new study by Kalshi, prediction markets beat Wall Street consensus estimates on US inflation with a 40% lower average error over a 25-month period. Moreover, the analysis shows that these markets were particularly accurate when inflation deviated sharply from economists’ expectations.

Comparing inflation forecasts on its platform with professional consensus, Kalshi found that market-based traders outperformed conventional economists and analysts throughout the 25 months reviewed. The performance edge was most visible during periods of heightened economic volatility, when traditional models tend to struggle.

Market-based estimates of year-over-year changes in the Consumer Price Index (CPI) recorded a 40% lower average error than consensus forecasts between February 2023 and mid-2025, the study reports. However, the gap widened further when actual CPI readings diverged strongly from expectations, with Kalshi’s forecasts beating consensus by as much as 67% in those instances.

Crisis Alpha and the value of disagreement

The study, titled “Crisis Alpha: When Do Prediction Markets Outperform Expert Consensus?”, also examined how disagreement itself can signal upcoming surprises. Specifically, it looked at the relationship between the size of the gap between Kalshi’s CPI estimate and Wall Street consensus, and the likelihood of a shock.

When Kalshi’s CPI estimate differed from the consensus by more than 0.1 percentage point one week before the official release, the probability of a significant deviation in the actual CPI reading rose to about 80%, versus a 40% baseline. That said, the paper cautions that the sample of large shocks over the period is still relatively small.

The authors argue that this pattern highlights the potential for market-based forecasting to serve as an early warning tool for policymakers and institutional investors. Rather than replacing existing models, the results suggest that markets can act as a complementary signal, especially during episodes of structural change or financial stress.

Information aggregation and the wisdom of the crowd

Unlike traditional forecasting, which often relies on a shared set of economic models and assumptions, Kalshi’s markets aggregate information from a diverse set of traders with direct financial incentives. This structure, the study notes, creates a “wisdom of the crowd” effect that can respond more quickly to new data and shifting narratives.

That said, the report emphasizes that the advantage is most evident when the forecasting environment becomes challenging. In the authors’ words, “when the forecasting environment becomes most challenging, the information aggregation advantage of markets becomes most valuable.” This is precisely when institutional decision-makers face the greatest risk of being blindsided by consensus-based errors.

Why prediction markets outperform consensus during times of stress may come down to how they process information. Traditional forecasters across banks and research shops often draw on similar datasets and models, which can limit adaptability when economic conditions move outside historical norms, the study suggests.

Incentives, liquidity and real-time pricing

Prediction market platforms, by contrast, reflect the views of individual traders drawing on sector-specific insights, alternative datasets and sometimes contrarian narratives. Moreover, traders on these markets have capital at stake and are rewarded or penalized solely on forecast performance, rather than on reputational or organizational considerations.

Institutional forecasters, the study notes, can face career and organizational constraints that discourage bold calls, even when the data might justify them. However, on a trading platform, the payoff structure favors aligning prices with the most probable outcome, irrespective of how unconventional that outcome may look to consensus economists.

The continuous nature of pricing on platforms like Kalshi also matters. Market prices update in real time as new information arrives, while Wall Street consensus estimates are typically fixed several days before official data releases. This lag can be especially costly when conditions are shifting quickly, such as during inflation shocks or policy surprises.

Growth of crypto-linked prediction platforms

Kalshi’s user base has expanded recently following the integration of its prediction platform into major crypto wallet Phantom. Earlier this month, the company raised $1 billion at an $11 billion valuation, underscoring growing investor interest in markets where users can bet on macroeconomic and political outcomes.

In October, rival platform Polymarket was reported to be in talks to raise funds at a valuation as high as $15 billion. Together, these developments suggest a rapid institutionalization of what began as a niche market, with both traditional finance and crypto-native investors now participating.

Moreover, earlier this year, research by a data scientist found that Polymarket was 90% accurate in predicting how events would occur one month out, and 94% just hours before the event. Still, the study noted that acquiescence bias, herd behavior and low liquidity can sometimes lead to overestimated probabilities for certain outcomes.

Limits and use cases for market-based signals

The Kalshi paper stresses that its findings do not imply that expert forecasts have become obsolete. Instead, it argues that prediction markets can act as a complementary tool for institutional risk management, monetary policy analysis and portfolio construction, particularly when historical relationships break down.

The authors acknowledge that the number of major inflation shocks in the sample is limited, which naturally constrains statistical power. However, they argue that the consistency of the performance gap across volatile episodes still points to a meaningful information advantage embedded in market prices.

Rather than calling for a wholesale replacement of traditional methods, the study suggests that institutional decision-makers should consider incorporating market-based signals alongside existing models. In periods of structural uncertainty, such blended approaches may improve inflation forecast accuracy and reduce the risk of being caught offside by unexpected macro data.

Overall, the analysis highlights how diverse, incentivized traders and real-time pricing can make platform-based forecasts a useful addition to the toolkit of economists, investors and policymakers navigating an increasingly complex macroeconomic landscape.