6.3 bp
mean absolute error for the Kalshi median and mode for headline CPI on release day, before the data
Bloomberg: 8.1 bp
RESEARCH / MARKET EVIDENCE
Markets can show how expectations change in real time.
Two recent studies examine real prediction market prices for inflation, interest rates, and elections. They show where these prices add information and where the results need careful interpretation.
Kuest editorial synthesis of external studies. This is not original empirical research by Kuest. Data and findings belong to the cited authors.
01 / MACROECONOMICS
Kalshi prices tracked CPI and Fed decisions
Diercks, Katz, and Wright studied Kalshi macro contracts and compared their forecasts with benchmarks including Bloomberg consensus, fed funds futures, and surveys from the Federal Reserve Bank of New York.
0.0 bp
mean absolute error for the median and mode of the fed funds rate on FOMC days in the study sample
Fed funds futures: 1.0 bp
2022–2025
period covered by the macro sample analyzed in the study
Forecast error on event day
This table restates values reported by the authors. Errors are in basis points; asterisks mark statistically significant differences from the benchmark, using Diebold–Mariano tests.
| Variable | Benchmark | Kalshi mean | Kalshi median | Kalshi mode |
|---|---|---|---|---|
| Headline CPI · MAE | Bloomberg 8.1 | 6.9 | 6.3* | 6.3* |
| Core CPI · MAE | Bloomberg 7.0 | 7.0 | 8.0 | 7.0 |
| Unemployment · MAE | Bloomberg 10.9 | 11.7 | 10.7 | 10.7 |
| Fed funds · MAE | Futures 1.0 | 1.0 | 0.0** | 0.0** |
MAE = mean absolute error. * p < 0.10; ** p < 0.05. The study finds no statistically significant difference for core CPI or unemployment. The reported zero for Fed median and mode applies to this sample and forecast horizon.
The added value is seeing the full distribution
Surveys tend to publish estimates at spaced intervals. In the contracts studied, prices and probabilities moved over time and responded to inflation, employment, and FOMC news. A distribution also shows uncertainty and tail scenarios, not just one central estimate.
02 / ELECTIONS
The Polymarket study is an interesting case with clear limits
Cutting and coauthors analyzed daily Polymarket prices and polling data during the 2024 U.S. presidential election. They report that the market favored Trump in five of seven swing states and reacted to campaign events.
This does not prove markets always beat polls
A contract price about who wins is an implied probability; a poll of voting intention measures preference among voters. Those are different quantities, so putting them on the same “chance of winning” scale requires caution. The paper also studies one election and one platform, notes uncertainty about who traded, and discusses the possibility that large positions affected prices.
The soundest use of this study at Kuest is to illustrate the frequency of prices and their response to news, not to claim universal superiority over election polls.
03 / WHAT THIS MEANS FOR OPERATORS
A new information layer for well-defined markets
Together, the studies show that contracts with verifiable outcomes can record expectations at high frequency and reveal scenarios that occasional surveys may miss. This can help operators organize markets around events their audiences care about.
- 01Choose questions with observable outcomes, clear rules, and a reliable resolution source.
- 02Show the price, update time, liquidity, and rules; explain that a market price is neither a representative poll nor a guarantee.
- 03Evaluate performance by category, period, liquidity, and metric. Kalshi’s macro results do not automatically validate other platforms or markets.
These studies did not evaluate Kuest infrastructure, liquidity, or forecast accuracy. They document evidence from third-party markets; they are not proof of Kuest product performance.
SOURCES AND DATA
Read the studies and explore available materials
FEDS · 2026-010
Federal Reserve · FEDS 2026-010
“Kalshi and the Rise of Macro Markets,” by Anthony M. Diercks, Jared Dean Katz, and Jonathan H. Wright. Published in February 2026 in the FEDS series.
DATA · REPLICATION
Replication data and code
The authors’ repository provides Kalshi data and code. Bloomberg and futures data used in some comparisons are not included; the README also notes a later change to Kalshi’s historical API.
ARXIV · 2507.08921
arXiv · 2507.08921
“Are Betting Markets Better than Polling in Predicting Political Elections?”, by Laurie E. Cutting and coauthors. Version 1 preprint submitted in July 2025.
Suggested citation
Diercks, A. M., Katz, J. D., & Wright, J. H. (2026). “Kalshi and the Rise of Macro Markets.” Finance and Economics Discussion Series 2026-010. DOI: 10.17016/FEDS.2026.010.
Page prepared in September 2026.
INFRASTRUCTURE FOR OPERATORS
Build a market around the questions your audience follows.
Kuest provides white-label infrastructure for organizations that want to launch markets for their brand and audience.