Predictive Market Platform Kalshi Releases First Research Report Revealing that its CPI Data Forecasts Have an Average Absolute Error 40% Lower Than Traditional Consensus Expectations, with Up to 60% Higher Accuracy During Major Market Shocks, Demonstrating the Advantages of “Collective Intelligence” in Economic Forecasting.
(Background: CNBC Collaborates with Prediction Market Kalshi, Real-Time Odds Fully Launching on TV Shows and News Reports)
(Additional Context: From Ballet Dancer to Youngest Female Billionaire: How Luana Built the Billion-Dollar Prediction Market Kalshi)

Table of Contents

• Overview
  • Key Highlights
• Background
• Methodology
  • Data
  • Shock Classification
  • Performance Metrics
• Results: CPI Forecast Performance
  • Overall Accuracy Advantage
  • “Shock Alpha” Indeed Exists
• Extended Discussion
  • Participant Heterogeneity and “Collective Wisdom”
  • Differences in Participant Incentive Structures
  • Information Aggregation Efficiency
• Limitations and Notes
• Conclusion

Editor’s Note:

The leading prediction market platform Kalshi announced yesterday the launch of a new research column, Kalshi Research, aimed at providing scholars and researchers interested in prediction market topics with internal Kalshi data. The first research report of this column is now published. Below is the original content of the report, translated by Odaily Planet Daily:

Overview

Typically, in the week prior to the release of major economic statistics, analysts and senior economists at large financial institutions will provide estimates of expected values. These forecasts are aggregated into what is widely known as the “Consensus Expectation,” which is considered an important reference for insights into market shifts and position adjustments.

In this research report, we compare the performance of consensus expectations and Kalshi’s implied pricing (hereafter sometimes referred to as “market forecasts”) in predicting the actual value of a key macroeconomic indicator—the Year-over-Year (YoY) CPI.

Key Highlights

· Overall Accuracy Advantage: Under all market conditions (including normal and shock environments), Kalshi’s predictions have an average absolute error (MAE) that is 40.1% lower than consensus expectations.

· “Shock Alpha”: During significant shocks (greater than 0.2 percentage points), within a one-week forecast window, Kalshi’s predictions outperform consensus MAE by 50%. When considering data released the day before, this advantage increases to 60%. During moderate shocks (0.1–0.2 percentage points), within the same forecast window, Kalshi’s MAE is also 50% lower than consensus; the day before data release, the advantage extends to 56.2%.

· Predictive Signal: When the market forecast deviates from consensus expectations by more than 0.1 percentage points, the probability of a shock occurring is approximately 81.2%, rising to about 82.4% the day before data release. In cases where market forecasts and consensus expectations diverge, market predictions are more accurate in 75% of cases.

Background

Macroeconomic forecasters face an inherent challenge: The most critical moments for prediction—market dislocations, policy shifts, and structural breaks—are precisely when traditional models tend to fail most. Financial market participants typically release consensus forecasts days before key economic data are published, aggregating expert opinions into market expectations. However, these consensus views, while valuable, often share similar methodologies and information sources.

For institutional investors, risk managers, and policymakers, the stakes of forecast accuracy are asymmetric. During stable periods, slightly better forecasts offer limited value; but during market turmoil—when volatility surges, correlations break down, or historical relationships fail—superior accuracy can generate significant Alpha and limit drawdowns.

Therefore, understanding how parameters behave during market volatility is crucial. We focus on a key macroeconomic indicator—YoY CPI—which is central to future interest rate decisions and a vital signal of economic health.

We compare and evaluate forecast accuracy across multiple time windows before official data releases. Our core finding is that the so-called “Shock Alpha” indeed exists—market-based forecasts can achieve additional predictive precision during tail events compared to baseline consensus. This outperformance is not merely academic; it significantly enhances signal quality at moments when forecast errors are most costly. In this context, the key question is not whether prediction markets are “always correct,” but whether they provide signals that are worth integrating into traditional decision-making frameworks due to their unique, differentiated value.

Methodology

Data

We analyze the daily implied forecasts from traders on the Kalshi platform, covering three time points: one week before data release (aligned with consensus forecast timing), the day before release, and the morning of release. Each market used is (or was) a genuinely tradable, live market reflecting real capital positions at different liquidity levels. For consensus expectations, we collected institutional-level YoY CPI forecasts, typically published about a week prior to the official release by the U.S. Bureau of Labor Statistics.

The sample period spans from February 2023 to mid-2025, covering over 25 monthly CPI release cycles across various macroeconomic environments.

Shock Classification

Events are categorized based on the relative “unexpected magnitude” compared to historical levels. “Shocks” are defined as the absolute difference between consensus expectations and actual data:

· Normal Events: CPI forecast error less than 0.1 percentage points;

· Moderate Shocks: CPI forecast error between 0.1 and 0.2 percentage points;

· Major Shocks: CPI forecast error exceeding 0.2 percentage points.

This classification allows us to examine whether forecast advantage varies systematically with forecast difficulty.

Performance Metrics

To evaluate forecast performance, we use:

· Mean Absolute Error (MAE): the primary accuracy metric, calculated as the average of absolute differences between predicted and actual values.

· Win Rate: when the difference between consensus and market forecast reaches or exceeds 0.1 percentage points (rounded to one decimal), we record which forecast is closer to the actual outcome.

· Forecast Horizon Analysis: tracking how the accuracy of market valuations evolves from one week before to the release date, revealing the value of continuous information incorporation.

Results: CPI Forecast Performance

Overall Accuracy Advantage

Across all market conditions, market-based CPI forecasts have an MAE that is 40.1% lower than consensus forecasts. At all forecast horizons, market-based predictions outperform consensus by 40.1% (one week ahead) to 42.3% (one day ahead).

Furthermore, when there is divergence between consensus expectations and implied market values, Kalshi’s market-based forecasts show statistically significant win rates, ranging from 75.0% one week ahead to 81.2% on the release day. Including cases where forecasts are effectively tied (to one decimal), the market-based predictions are equal or better than consensus in about 85% of cases one week prior.

Such a high directional accuracy indicates: when market forecasts diverge from consensus, this divergence itself carries significant information about the likelihood of shocks.

“Shock Alpha” Indeed Exists

Forecast accuracy differences are especially pronounced during shock events. During moderate shocks, when release timing is aligned, market forecasts’ MAE is 50% lower than consensus; this advantage widens to 56.2% or more the day before data release. During major shocks, the market forecast MAE is also 50% lower when release timing is aligned, reaching 60% or more the day before. In normal, non-shock environments, market and consensus forecasts perform similarly.

Although the sample size for shocks is smaller (which is reasonable given the inherently unpredictable nature of shocks), the overall pattern is clear: when the environment is most challenging to forecast, the information aggregation advantage of the market is most valuable.

More importantly, not only does Kalshi’s forecast outperform during shocks, but the divergence between market and consensus forecasts itself may serve as a signal of impending shocks. When divergence exceeds 0.1 percentage points, the probability of a shock is about 81.2%; the day before data release, this probability rises to approximately 84.2%.

This practically significant difference suggests that prediction markets are not only competitive forecasting tools alongside consensus but also serve as a “meta-signal” of forecast uncertainty, transforming divergence into an early warning indicator of potential surprises.

Extended Discussion

An obvious question follows: Why do market forecasts outperform consensus during shocks? We propose three complementary mechanisms to explain this phenomenon.

Participant Heterogeneity and “Collective Wisdom”

While traditional consensus forecasts aggregate multiple institutional viewpoints, they often share similar methodological assumptions and information sources. Econometric models, Wall Street research reports, and government data releases form a highly overlapping knowledge base.

In contrast, prediction markets aggregate positions held by participants with diverse information bases: proprietary models, industry insights, alternative data sources, and experiential intuition. This diversity has a solid theoretical foundation in the “wisdom of crowds” concept. It suggests that when participants possess relevant information and their forecast errors are not perfectly correlated, aggregating independent predictions from varied sources tends to produce more accurate estimates.

During “state shifts” in macro environments, this diversity becomes especially valuable—fragmented, localized information held by individuals interacts within the market, combining into a collective signal.

Differences in Incentive Structures

Institutional consensus forecasters often operate within complex organizational and reputational systems that systematically deviate from “pure” accuracy pursuit. The professional risk of large forecast errors creates an asymmetric reward structure—significant reputational costs for errors, while highly accurate forecasts, especially those deviating from the consensus, may not be proportionally rewarded.

This asymmetry induces herding behavior—forecasters tend to align their predictions close to the consensus, even if their private information or models suggest otherwise. The reason is that, within their career systems, “being wrong alone” is costly, whereas “being right alone” is less incentivized if it involves diverging from the crowd.

In stark contrast, prediction market participants face a direct alignment of forecast accuracy with economic outcomes—correct predictions lead to profits, errors lead to losses. Reputation effects are minimal; the only cost of diverging from consensus is financial loss, which depends solely on correctness. This structure exerts stronger selection pressure for accurate forecasting—participants who can systematically identify and exploit consensus errors can accumulate capital and influence. Conversely, those who merely follow the consensus risk losses when it is wrong.

During periods of heightened uncertainty, when institutional forecasters’ divergence from the consensus becomes most costly, this incentive structure divergence is most pronounced and economically significant.

Information Aggregation Efficiency

A notable empirical fact is that even one week before data release—aligned with the typical timing of consensus forecasts—market predictions still outperform. This indicates that the market’s advantage is not solely due to faster access to public information.

Instead, market forecasts may more efficiently aggregate dispersed, industry-specific, or fuzzy information that is difficult to formalize into traditional econometric models. The relative edge of prediction markets may lie in their ability to synthesize heterogeneous, fragmented signals within the same time window, rather than simply “getting there earlier.” Traditional consensus mechanisms, even with the same information timing, often struggle to process and combine such diverse data efficiently.

Limitations and Notes

Our findings are subject to important limitations. The sample covers only about 30 months, and major shocks are inherently rare, limiting statistical power for tail events. Longer time series would improve inference, though current results strongly suggest the superiority and signal value of market forecasts.

Conclusion

We document the significant and economically meaningful outperformance of prediction markets relative to expert consensus, especially during shocks. Market-based CPI forecasts are approximately 40% more accurate overall, with reductions in error reaching about 60% during major structural shifts.

These findings point to future research directions: expanding sample sizes across multiple macro indicators, exploring whether “Shock Alpha” events can be predicted via volatility and divergence metrics; identifying liquidity thresholds where prediction markets reliably outperform traditional methods; and examining the relationship between market implied forecasts and high-frequency trading-based implied predictions.

In environments heavily reliant on correlated models and shared information, prediction markets offer an alternative information aggregation mechanism capable of early detection of regime shifts and more efficient processing of heterogeneous signals. For decision-makers operating in increasingly uncertain, structurally volatile, and tail-risk-prone economic environments, “Shock Alpha” may not only reflect incremental forecasting improvements but also serve as a fundamental component of robust risk management infrastructure.