Kalshi’s 25-month review argues that prediction markets produced materially more accurate inflation forecasts than Wall Street consensus from February 2023 through mid-2025. The analysis reports a 40% lower average error for year-on-year CPI, with the performance gap widening to as much as 67% when inflation outcomes deviated sharply from expectations.
Why Kalshi says prediction markets outperformed consensus
Kalshi compared market-based forecasts on its platform with traditional consensus estimates for year-on-year CPI across a continuous 25-month window. The study’s headline result is a 40% reduction in average forecasting error for prediction-market prices versus Wall Street consensus, and it highlights that during high-volatility episodes the relative advantage reached up to 67%. The work is framed around realized forecasting error rates, rather than model diagnostics or holdout testing.
Kalshi attributes the reported outperformance to how market pricing works in practice. Continuous trading is presented as a mechanism that updates prices in real time up to settlement, which the study contrasts with analyst consensus statements that are often fixed before official releases. It also argues that aggregation across many participants captures heterogeneous information sets, producing a “wisdom of the crowd” signal.
The review also emphasizes incentives as a differentiator. Kalshi’s interpretation is that direct financial incentives push participants to trade on the most probable outcomes, whereas institutional forecasting may face constraints that are not purely accuracy-driven. In this framing, responsiveness to evolving volatility comes from the combination of real-time repricing, diverse participation, and incentive alignment.
Kalshi positions these market signals as a complementary input for institutions. The study argues prediction-market prices can support risk management, policy analysis, and portfolio construction when used alongside traditional forecasting frameworks, particularly in volatile environments where consensus may lag. At the same time, it flags practical constraints, noting that regulatory hurdles and legal challenges could affect scalability and institutional adoption.
Overall, the analysis advances the view that market-priced probabilities can deliver stronger short-term inflation forecasting performance than conventional consensus estimates over the period studied. The implied operational takeaway is that institutions may increasingly integrate market signals into forecasting workflows, while accounting for compliance and adoption constraints.
