727 FinanceThe core value of prediction markets is producing high-quality probabilistic data, which has been adopted by the Federal Reserve and Wall Street financial institutions as a reference for macroeconomic decision-making. This article analyzes their price advantages, application scenarios, and data quality, revealing how on-chain prediction platforms are becoming the next generation of decision-making infrastructure.
Written by Ryan Yoon, Tiger Research
Compiled by Chopper, Foresight News
Our previous two articles introduced the definition and operating mechanism of prediction markets, as well as the regulatory gaps facing prediction markets in Asia. Our core conclusion is that the core value of prediction markets is producing data, not merely serving as a trading venue.
Major financial media and market infrastructure providers have begun adopting prediction market data, which demonstrates the value of such data. Google Finance, CNBC, CNN, and The Wall Street Journal have begun displaying prediction market probabilities, while Intercontinental Exchange (ICE), the parent company of the New York Stock Exchange, and bond trading platform Tradeweb now also distribute prediction market data to institutional clients.
Adoption timeline of prediction market data, November 2025 — June 2026
A Federal Reserve working paper also treats prediction market data as a benchmark for macroeconomic expectations, used to assess indicators such as interest rate decisions, inflation, unemployment, and GDP growth. A Federal Reserve economist commented on prediction markets such as Kalshi: "High-frequency, continuously updated, and rich in distributional information, they are highly valuable to both researchers and policymakers." Prediction markets are increasingly being viewed as data sources rather than betting websites.
The industry has recognized the data value of prediction markets and put it into practical use. This article will examine three questions to break down the practical utility of such data:
Why can prediction market transaction prices become valuable data?
In what scenarios is prediction market data applied?
Can current prediction markets produce qualified, high-quality data?
Why can prediction market prices become high-value data?
As noted in our previous article, the most basic instrument in prediction markets is a binary contract: it pays $1 if the event occurs and $0 if it does not. For example, a "rain tomorrow" contract trading at $0.62 means market participants judge the probability of the event occurring to be 62%.
Therefore, the prediction market price ($0.62) can be directly converted into the probability of a future event occurring (62%). The four underlying characteristics below explain why these prices can be incorporated by global financial institutions and media into real decision-making references, rather than being merely betting odds.
Four major characteristics that make prices valuable data
Skin in the game ensures credibility
Many people simply equate prediction market trading with gambling, but putting one's own money at stake means traders truly have skin in the game.
Skin in the game means traders must bear risk and consequences for their judgments. Expressing an opinion casually costs nothing, but once a prediction is wrong, traders lose real money. Therefore, traders will only reflect verified information and genuine judgments in their quotes.
When judgments backed by large amounts of capital converge, as long as market liquidity is sufficient, prediction market prices will form highly credible probability data.
Continuous real-time updates
Traditional polls and expert surveys only capture opinions at the moment of the survey and struggle to quickly capture sudden changes. For example, before a central bank interest rate decision, if an economic shock or major announcement occurs, polls completed days earlier or expert surveys updated monthly cannot reflect the new changes at all.
In contrast, once new information enters the market, prediction market prices change immediately. When news, sudden developments, or regulatory announcements appear, traders adjust positions, driving continuous price changes.
Prediction market probabilities are not static numbers frozen in the past, but real-time data that changes in sync with developments.
Coverage of all outcomes, outputting a complete probability distribution
Most traditional indicators provide only a single average or most likely value, compressing multiple views into a single number and losing a great deal of information. For example, in surveys on central bank rate cut expectations, expert questionnaires often give only one result—"cut rates by 0.25 percentage points"—and cannot reflect different scenarios arising from changes in committee members' statements.
Prediction markets, by contrast, can output the probability distribution corresponding to all outcomes, rather than giving only a single conclusion. For example: keep rates unchanged 20%, cut rates by 0.25 percentage points 70%, cut rates by 0.5 percentage points 10%, and this set of probabilities is a continuous time series.
It can not only measure the most likely baseline scenario, but also capture tail risks—that is, scenarios with low probability but enormous impact if they materialize. Therefore, prediction markets do not merely provide a single-point expectation; the complete probability distribution can help assess overall market volatility and potential risks.
Ex-post verifiability
The most important reason academia and global financial markets trust prediction market data is that results can be fully verified ex post.
Although ordinary expert forecasts and polls can also be reviewed afterward, the evaluation criteria leave considerable room for interpretation, and institutions often cite only historical records that favor them. Prediction market data, however, are clear, quantified probability records; settlement outcomes are only 0 or 1, and the evaluation criteria are fully mathematical, with no room for subjective interpretation.
Anyone can transparently compare historical prices with actual outcomes to verify accuracy. This objective, tamper-proof historical record allows data practitioners and institutions to rigorously evaluate past performance, which is the reliable foundation for its use as a core decision-making indicator.
Application scenarios of prediction market data
Prediction markets have not been in the public eye for very long, but their earliest origins can be traced back to the Iowa Electronic Markets (IEM) in the 1980s. For a long time, their most well-known use was election forecasting; in recent years, application scenarios have rapidly expanded, further highlighting their value.
Election forecasting: earlier and more accurate than polls
The Iowa Electronic Markets (IEM) was the first case to demonstrate the potential of prediction markets. Research comparing 964 polls across five U.S. presidential elections from 1988 to 2004 found that in 74% of cases, IEM forecasts were more accurate than polls.
Comparison of IEM forecasts and polls in U.S. presidential elections
This advantage does not appear only on the eve of elections. Analysis of the 1988–2000 elections shows that one week before the election, the IEM's mean forecast error for vote share was 1.5 percentage points, lower than the 2.1 percentage point error of Gallup's final poll; even 150 days before the election, the average error was only about 5 percentage points.
However, the two sets of data cannot be directly equated: polls answer "what would the result be if the election were held today," a snapshot of the present; prediction market prices forecast "the final result on election day," a forward-looking expectation.
Some studies also point out that polls adjusted for the remaining election time can be more accurate than prediction markets; but another analysis of the 2008 election showed that after correcting for biases inherent in prediction markets, prediction markets outperformed adjusted polls during the early campaign period and in close races.
In summary, prediction markets are more accurate than unadjusted raw polls and comparable to finely calibrated polls; the additional value is that months before the election, they can continuously update forecasts daily.
Macroeconomics: real-time reading of market expectations
For macroeconomic indicators such as interest rates, inflation, and employment, it is crucial to grasp true market expectations before official release.
Traditional methods rely on expert surveys or traditional financial product prices. Surveys are updated only every few weeks, lacking timeliness; futures prices provide only a single expected value, and option-implied distributions cover only a few indicators such as interest rates and headline inflation. Prediction markets, by contrast, can directly generate real-time probability distributions through contract prices corresponding to different outcomes.
This quantitative value was already validated in the economic derivatives market from 2002 to 2005. Research on binary options for economic data such as nonfarm payrolls, retail sales, and initial jobless claims found that market-implied expectations were as accurate as or more accurate than the average of expert surveys, and that the probability distributions given by the market closely matched the actual frequency of events.
Derivatives market forecasts vs. expert survey forecasts (2002–2005)
A recent Federal Reserve working paper compared Kalshi prediction market forecasts with mainstream benchmarks: one day before FOMC meetings, Kalshi's median and modal forecasts for the federal funds rate were exactly correct, significantly outperforming federal funds rate futures. Compared with the Bloomberg consensus, Kalshi's forecast accuracy for headline CPI inflation was significantly higher, while its forecast accuracy for core CPI inflation and unemployment was statistically similar to the Bloomberg consensus. These findings indicate that prediction markets can outperform certain benchmarks in some respects while performing comparably in others.
The research also mentions broader practical value: key macroeconomic indicators such as GDP growth, core inflation, unemployment, and nonfarm payrolls previously lacked liquid derivatives indicators. Prediction markets fill this gap by providing real-time market-based expectations and detailed probability distributions, serving as a reference benchmark for researchers and policymakers.In summary, for indicators such as overall CPI, prediction markets are more accurate than traditional indicators; for interest rate decisions, they are at least on par with traditional tools, and at the same time they can output real-time probability distributions of macroeconomic expectations that traditional financial markets cannot cover, giving them strong practical value.
Public health: influenza warnings 2–4 weeks in advance
The U.S. Centers for Disease Control and Prevention (CDC) publishes official influenza statistics weekly, so there is a time lag in tracking the actual spread of influenza.
To solve this problem, researchers at the University of Iowa built a pilot influenza prediction market in which healthcare workers participate in trading. The market aggregates clinical information voluntarily reported by healthcare practitioners and can predict the intensity of influenza activity in Iowa 2–4 weeks in advance.
Such forecasts can assist in scheduling diagnosis and treatment, epidemic prevention measures, and public policy making. This case demonstrates that prediction markets can support rapid public health response, although the pilot was limited to Iowa.
Enterprises and intelligence agencies: bypassing reporting hierarchies to obtain frontline information
In conventional corporate structures, negative news and uncertain expectations tend to be diluted or glossed over as they are reported upward level by level. Internal corporate prediction markets, by contrast, let employees bet directly on internal events such as revenue, new product launch dates, and hiring targets, with rewards for correct guesses. They bypass rigid reporting processes and quickly aggregate frontline judgments and information into prices.
HP: an internal sales forecasting market had 20–26 active traders; in 8 comparable cases, its forecasts were more accurate than management's official expectations 6 times, and in all 8 cases it correctly identified the direction of deviation in official expectations.
Intel: across 979 internal prediction markets run from 2006 to 2013, more than half were more accurate than the company's official forecasts.
Ford: in 2011, based on trading signals from an internal feature prediction market, the company decided not to offer Ford-brand bicycle racks or in-car vacuum cleaners on subsequent models.
Google: executives mentioned at an all-hands meeting that a target for hiring for an executive position had not been met for six consecutive quarters, a result the internal prediction market had anticipated long before.
Siemens: a traditional schedule for a software project showed on-time delivery, but the internal prediction market warned that the project would be delayed by 2–3 weeks; the project ultimately finished 13 days late.
Prediction markets are also useful for geopolitical and national security assessments. The U.S. Intelligence Advanced Research Projects Activity (IARPA) held a geopolitical forecasting tournament from 2011 to 2015 comparing prediction markets with forecasting surveys (forecaster questionnaires): across two experimental rounds, more than 2,400 participants, and 261 events in total, prediction markets were more accurate than simple survey averages but less accurate than a statistically adjusted team survey weighted by historical accuracy.
In corporate settings, multiple cases demonstrate that prediction markets can capture information missed by formal reporting and official forecasts. In the intelligence tournament, prediction markets outperformed simple questionnaire averages but were weaker than weighted, optimized team surveys. At the same time, cases such as HP and Siemens involved only 20–60 traders, a small sample, which is a clear limitation.
Can today's prediction markets produce high-quality data?
Can prediction markets currently in operation produce this kind of valuable data? Based on public API data from the on-chain prediction platform Limitless through October 2026, we analyzed market types and forecast accuracy.
A rich variety of tradable categories
At the time of analysis, the platform had 556 active market instruments in total, far beyond simple sports and entertainment events, also covering policy and macroeconomic indicators such as major central bank interest rate decisions, elections, geopolitical variables, tech company IPOs, and quarterly earnings.
Prediction market instruments have expanded from a small number of events to a full range of macroeconomic and policy variables. The accuracy, timeliness, and verifiability verified above are no longer confined to a few domains and are extending to global economic and policy decision-making scenarios.
Uncertainties and future volatility that are difficult for traditional finance and polling to quantify can now be converted into credible probability data for traders and policymakers to use as decision-making references.
Forecast accuracy
To verify whether market prices match the true probability of events occurring, we selected 2,681 settled instruments (about 85% of which were sports markets) and compared contract prices one day before settlement with the final actual outcomes.
Contract pricing one day before settlement versus the true probability of the event occurring
Across 10 price bands, the gap between actual occurrence frequency and market pricing was kept within 4 percentage points in 9 bands. For example, among 165 contracts priced in the 70%–80% band one day before settlement, the final realized proportion was 74.5%; among 113 contracts priced above 90%, the realized proportion was 97.3%.
The only band with a notable error was 50%–60% (realized rate 48.8%), mainly because about one-third of the instruments in that band (mostly sports events) remained at the initial listing price of 50% until one day before settlement, with very poor liquidity.
Excluding such low-liquidity, low-attention sports instruments, prediction market prices can reflect the true probability of events occurring with high precision.
Responsiveness to new information
Timeliness underpins accuracy. We tracked the price time series of contracts related to Federal Reserve interest rate decisions to verify whether the market can quickly digest new information.
For Federal Reserve interest rate decision instruments, probabilities adjusted immediately whenever new economic data or policy news was released. In mid-August, the probability of keeping rates unchanged was about 70%; from late August to early September, as major indicators were released, that probability fell below 50%; after the inflation data release on September 11, the probability of holding rates steady fell to about 20%; and one day before the decision it fell further to the 10% range. Correspondingly, the probability of a 0.25 percentage point rate hike rose to 86%–89% on the day of the decision, broadly in line with the roughly 90% expectation implied by CME FedWatch federal funds rate futures. Other major central bank decision instruments on Limitless also exhibit this real-time responsiveness.
Limitless platform September 2026 Federal Reserve meeting event market: probability trends for different outcomes
So on-chain prediction platforms today already meet the timeliness and accuracy required for high-quality data. These cases demonstrate that prediction market prices can absorb new information in real time and continuously output high-precision probability data.
Do prediction markets really have practical value?
The core value of prediction markets does not lie in speculative trading or earning returns. Long-term academic research, combined with the empirical analysis in this article, shows that in scenarios such as elections, macroeconomic indicators, and corporate expectations, the probability data output by prediction markets is more accurate and timely than traditional surveys and expert forecasts, or serves as an effective supplement. The fact that global financial media and core trading infrastructure have integrated prediction market data rests precisely on this real-world utility.
In many Asian countries and other regions, prediction markets are still generally regarded as gambling. But this analysis shows that platforms such as Limitless have already listed instruments on major regional policy and economic events, including Bank of Korea and Bank of Japan interest rate decisions, although liquidity in such instruments remains relatively weak.
For such data to become trusted decision-making infrastructure globally, including in Asia, three things need to be done:
First, reduce price distortion and prevent manipulation through mechanism design. Improve liquidity in markets for major macroeconomic and political events, expand the base of rational traders, and avoid small amounts of capital or speculators distorting prices.
Second, proactively engage with regulators to change the entrenched policy-level perception of prediction markets. Continuously engage with regulators in major Asian economies, build cooperation frameworks, and encourage regulators to view prediction markets as an auxiliary macro-monitoring tool that aggregates collective public information, rather than merely an object to be controlled or banned.
Third, maintain transparent and open data and cultivate an independent research ecosystem. Regularly publish accuracy metrics and time-series datasets, support research by specialized analysis institutions such as Limitless Research, and make it easy for research institutions, media, and policy departments to access data, building trust on the basis of objective evidence.
Prediction market platforms are at a turning point, evolving from mere trading venues into the next generation of data infrastructure. If they can continuously and transparently output and verify real-time probability data covering global (including Asian) policy and economic events, they will become an objective and practical decision-making tool in an environment of uncertainty.