Prediction markets are moving beyond election-night speculation and into the financial information pipeline, with their live contract prices increasingly used as probability estimates for interest-rate decisions, inflation releases, public-health trends and political outcomes. Google Finance, CNBC, CNN and The Wall Street Journal now display prediction-market odds in some coverage, while Intercontinental Exchange and Tradeweb have distributed related data alongside established market datasets for professional clients.
The appeal is straightforward: a contract that pays $1 if an event occurs and nothing if it does not can translate its price into an implied probability. A contract trading at $0.62, for example, indicates that the market assigns roughly a 62% chance to the outcome, before accounting for any trading frictions or market-specific biases.
A Federal Reserve Board working paper by Diercks, Katz and Wright in 2026 treated prediction-market prices as a potential benchmark for expectations around policy rates, inflation, unemployment and gross domestic product. Its findings place platforms such as Kalshi alongside more familiar gauges including economist surveys and interest-rate futures, though their usefulness varies by event, liquidity and the amount of information available to traders.
Prices offer a live view of changing expectations
Prediction markets differ from polls and periodic forecasts because their prices can move continuously as new information arrives. They can also offer contracts across multiple possible outcomes, producing a distribution of expected results rather than a single headline estimate.
The structure rests on four practical features described in the supplied research: participants have money at risk, prices update as views change, markets can represent several possible outcomes, and contracts can be checked against a final result after settlement. That last feature makes calibration measurable. If contracts priced around 70% settle successfully around seven times out of 10 over a large enough sample, the market is broadly well calibrated.
The Federal Reserve study found that Kalshi’s median and mode forecasts for the federal funds rate were correct one day before Federal Open Market Committee decisions, outperforming fed funds futures at that short horizon, according to Diercks, Katz and Wright. The paper also reported stronger results for Kalshi than Bloomberg consensus estimates on headline consumer-price inflation, with statistically comparable performance on core CPI and unemployment.
One example tracked a Federal Reserve rate-decision market through late summer and early autumn. The implied chance of no rate change stood near 70% in mid-August, slipped below 50% after macroeconomic releases in late August and early September, then dropped to around 20% after Sept. 11 inflation data. It approached 10% immediately before the decision, while the probability of a 25-basis-point cut rose to between 86% and 89%. That was close to the roughly 90% probability implied by CME FedWatch, as cited by Diercks, Katz and Wright and Daiwa Capital Markets.
Election research shows useful, but uneven, results
Academic evidence on political forecasting has produced a more nuanced record than the popular view that markets always beat polling.
Research by Joyce Berg, Forrest Nelson and Thomas Rietz covering five US presidential elections from 1988 through 2004 found that Iowa Electronic Markets forecasts outperformed 74% of 964 conventional poll comparisons. A separate review by Kenneth Arrow and co-authors found that IEM’s average vote-share error one week before Election Day was 1.5 percentage points, compared with 2.1 percentage points for Gallup’s final poll over the 1988-2000 period.
The gap was larger further from Election Day. About 150 days before voting, IEM’s error was close to 5 percentage points, according to Arrow and co-authors. That result reflects a central limitation for all election forecasting tools: early prices and polls must incorporate a long period in which campaigns, economic conditions and candidate perceptions can change.
Polls and markets also answer different questions. Polling generally measures voter preferences at the time respondents are surveyed. A market price incorporates traders’ assessment of how those preferences, turnout and later developments could affect the final Election Day result. Robert Rothschild’s 2009 review of the 2008 election found that adjusted prediction-market performance exceeded adjusted polling in earlier campaign stages and close contests, after accounting for known market biases.
Ericson and Wlezien’s 2008 work also found that time-adjusted polling could be marginally more accurate than prediction markets. The record suggests that prediction prices work best as a complementary expectation measure rather than as a replacement for survey data.
Onchain markets face a liquidity test
An analysis of the onchain prediction platform Limitless, using public API data through October 2026, examined 2,681 settled event contracts and found broadly close alignment between prices one day before settlement and final outcomes. Roughly 85% of the sample consisted of sports markets, a concentration that limits how directly the findings can be applied to economics or politics.
Across 10 price ranges, nine showed a difference of no more than four percentage points between the contract’s implied probability and its realized settlement frequency. In the 70% to 80% range, 165 contracts produced a realized success rate of 74.5%. Among 113 contracts priced at 90% or above, 97.3% settled in the predicted direction.
The largest gap appeared in the 50% to 60% bucket, where the realized frequency was 48.8%. The analysis linked much of that distortion to thin trading: about one-third of these contracts, mainly niche sports listings, remained at their initial 50% price one day before settlement.
That detail is critical for anyone treating onchain contract prices as data. A quoted probability reflects the market’s information only if participants are actively trading and prepared to move the price. An untouched 50-cent contract may represent a listing default rather than a considered assessment of the event.
Limitless listed 556 active markets during the sampling period, spanning macroeconomic releases, policy decisions, elections, geopolitical developments and corporate events such as IPO processes and earnings results. The breadth makes onchain markets useful for observing niche events that may receive little coverage from conventional forecasting providers, but it also means liquidity can vary sharply from contract to contract.
Corporate and public-health experiments preceded today’s data feeds
Companies were testing internal prediction markets years before contract prices began appearing in mainstream financial coverage. At Hewlett-Packard, research by Kay-Yut Chen and Charles Plott found that employee markets beat management’s official sales forecast in six of eight comparable cases, while correctly anticipating the direction of management’s forecast error in all eight.
At Intel, 979 internal prediction markets operated between 2006 and 2013. More than half produced forecasts more accurate than official expectations, according to research by Gillen, Plott and Shum. Siemens also used an internal market that anticipated a two- to three-week delay on a software project; the project ultimately shipped 13 days late, according to Ortner’s 1998 account.
Public-health research has produced similar experiments. A University of Iowa pilot led by Philip Polgreen, Forrest Nelson and George Neumann used trading by medical professionals to forecast influenza activity in Iowa two to four weeks ahead of official weekly data from the US Centers for Disease Control and Prevention.
The expanding distribution of prediction-market prices gives traders, policymakers and journalists another real-time measure of expectations. Their strongest use is likely to be alongside polls, surveys, futures and official data, especially where contract liquidity is sufficient and the event’s settlement rules are clear.
Want to trade event outcomes directly? Explore Toobit’s event contracts and turn your macro or election views into positions.
Disclaimer: The content on this page is provided for general informational purposes only and does not represent the views or financial advice of Toobit. We make no guarantees regarding the accuracy or completeness of this information and shall not be held liable for any errors, omissions, or outcomes resulting from its use. Investing in digital assets involves risk; users should independently evaluate their financial situation and the risks involved. For further details, please consult our Terms of Service and Risk Disclosure.
