How economic-data markets work
How CPI, jobs-report, FOMC and recession markets work: what the contract asks, which number settles it, and what happens when the data is revised.
Updated November 10, 2026 · Lumina Markets
On this page
An economic-data market asks a narrower question than "is inflation a problem": will a specific number, published by a named agency on a stated date, land above a stated threshold?
Moving from a topic to a threshold is what allows the question to settle, and it puts the weight on the series and the source rather than on the subject.
What the contract asks
Most economic-data contracts take one of three shapes:
- Threshold. Will the published figure be above or below a stated level? Will CPI year-over-year print above 3.0%?
- Range. Will the figure land inside a stated band? These are often listed as a ladder of adjacent contracts covering a span of outcomes.
- Date or event. Will something happen by a stated date? Will the target range be lowered at or before the next scheduled meeting?
Each is binary: yes or no, priced on Lumina between 2¢ and 99¢, paying $1 or nothing at resolution if held that long. A 34¢ quote corresponds to an implied probability of roughly 34%.
Common release families
Several US statistical releases recur often enough to anchor their own contracts. Four of the most frequently referenced:
CPI: the inflation print
The Consumer Price Index is published monthly by the Bureau of Labor Statistics, covering the prior month. A contract can reference the year-over-year change or the month-over-month change, headline or "core" (excluding food and energy), and seasonally adjusted or not, all of which are different series from the same release (opens in new tab). A contract naming one doesn't settle on another. See CPI print.
The jobs report
The BLS Employment Situation report, typically released on the first Friday of the month. Nonfarm payroll change is the headline figure; the unemployment rate, labor-force participation and average hourly earnings come from the same release and can move in different directions. A contract naming payrolls settles on payrolls, whatever the other series did.
FOMC decisions and the fed funds target
The Federal Reserve's Federal Open Market Committee holds eight regularly scheduled meetings a year (opens in new tab), plus additional meetings as needed, and sets a target range for the federal funds rate. Changes are conventionally made in quarter-point steps (25 basis points), though the convention isn't a rule. Contracts here are usually date-and-direction: will the range be lowered, raised or held at a given meeting, and by how much. Because unscheduled action is possible, the contract's date language matters more than an assumption about "the next meeting." See FOMC and the fed funds rate.
NBER recession dating
US recessions are dated by the National Bureau of Economic Research's Business Cycle Dating Committee, which weighs depth, diffusion and duration (opens in new tab) across a range of indicators rather than applying the widely-quoted "two consecutive quarters of falling GDP" rule. Recession contracts define a recession in their own rules, for example NBER's dating or two consecutive negative quarters of real GDP. The committee's approach is explicitly retrospective: it "tends to wait to identify a peak until a number of months after it has actually occurred," so a contract that names NBER's dating can turn as much on the timing of the announcement as on the economy, and one that uses GDP turns on those releases instead. See NBER recession.
Where the settling number comes from
For a data market the resolution source does most of the work, because it determines which of several published numbers the contract actually settles on.
A fully specified data contract generally names four things:
- The agency and the exact series
Not "inflation" but which index, on which basis, seasonally adjusted or not
- The release
Which scheduled publication settles it
- The vintage
The first published figure, or a subsequently revised one
- The fallback
What happens if that release doesn't appear on schedule
The third is easy to overlook, and it's where two contracts referencing "the same number" can stop being the same contract. Any specific contract's terms govern, so the terms are the thing to read rather than the general pattern described here.
When the number changes after publication
Several of these series are revised after first release, which is why the vintage a contract names matters.
Payroll estimates are revised twice, then benchmarked. BLS publishes a first preliminary figure roughly three weeks after the reference period, then a second preliminary estimate the following month and a final sample-based estimate the month after that, two revisions before the figure holds (opens in new tab). Separately, an annual benchmark aligns the survey to more comprehensive unemployment-insurance records, with the final benchmark published each February alongside the January report. A first print that clears a threshold can be revised back below it, and whether that affects a contract depends on the vintage it names.
- About 3 weeks afterFirst preliminary estimate
- 1 month laterSecond preliminary estimate
- 2 months laterFinal sample-based estimate
- Every FebruaryAnnual benchmark revision
Seasonally adjusted CPI is revised on a five-year window; the unadjusted index isn't. BLS updates the seasonal factors each February and applies them to the previous five years (opens in new tab) of seasonally adjusted data; seasonally adjusted figures older than that are treated as final. The not-seasonally-adjusted index is not part of that revision cycle, which is why BLS recommends the unadjusted series for escalation contracts and advises against the seasonally adjusted one. A contract referencing the seasonally adjusted series is referencing a number that can still change.
Methodology changes. Weights and definitions are updated on published schedules, so a long-dated contract can outlive part of the methodology behind its own reference series.
And sometimes the release doesn't appear. A lapse in appropriations can suspend federal statistical publication, delaying or canceling releases including CPI and the jobs report. That is the scenario a contract's fallback language addresses, and it's one reason data markets and government shutdown markets can move together.
Units
Some of the confusion around threshold contracts is arithmetic rather than economics.
- A basis point is 0.01 percentage points. 25 basis points is 0.25 percentage points; a 50 bp move is half a percentage point.
- Percentage points are not percent. An unemployment rate moving from 4.0% to 4.2% rose 0.2 percentage points, which is also a 5% increase in the rate. A threshold stated in percentage points refers to the first.
- Rounding can decide a threshold. Where a published figure lands on the threshold itself, the outcome depends on the precision the agency publishes at and on how the contract defines "above", both of which are set by the contract's own terms.
How prices behave around a release
Data markets often move through a rough sequence, though none of this is guaranteed on any individual contract.
- Before the releaseDrifts on related news, or sits still
The quote can move on related releases, revisions to prior months and commentary, and can also sit unchanged for days.
- At the releaseMoves quickly toward 0¢ or $1
Once the figure publishes, the quote typically moves quickly toward one boundary of the price range, since the outcome is largely determined at that point.
- After the releaseSettles on the contract's terms
Settlement follows the contract's terms rather than the news. For a revision-sensitive series or a committee-dated question, that can come well after the figure is public.
That's observed behavior, not a strategy. There's no forecast here about any release and no view on where any figure will land. A contract that resolves against you pays zero, and a correct view on the economy paired with the wrong vintage or a misread settlement term still resolves against you.
Lumina Markets Inc. is a CFTC-registered Introducing Broker and Member NFA (NFA ID: 0576536). Trading event contracts involves substantial risk of loss and is not appropriate for all.
Where the mismatches occur
Each row below is an illustration of the same underlying gap: a reasonable reading of the subject diverging from the contract's actual terms. The examples are hypothetical, not descriptions of specific contracts.
| Gap | An illustration of how it arises |
|---|---|
| Series | A contract settles on headline CPI while the reader was following core CPI. Both appear in the same release, and they can move in different directions. |
| Vintage | The first payroll print clears a threshold; the second-month revision brings it back below. Which one settles depends on the vintage the contract names. |
| Date language | A contract references a specific meeting or window rather than "the next decision", so an unscheduled or rescheduled action falls outside it. |
| No release | An appropriations lapse delays publication past the contract's window, and the fallback terms rather than the data decide the outcome. |
| Rounding | The published figure lands on the threshold itself, so the outcome turns on published precision and on how the contract defines "above". |
| Committee timing | Economic conditions turn, but an NBER dating announcement arrives months later, potentially outside a contract's window. |
Reading the settlement terms alongside the question is what makes these visible in advance. It doesn't eliminate them: revisions, delays and rescheduled decisions are outside any participant's control.
Related reading
FAQ
How do economic-data prediction markets work?
Each is typically a binary contract on whether a specific published figure clears a stated threshold, lands in a stated range, or arrives by a stated date. It pays $1 if yes and nothing if no, and settles on the figure published by the source the contract names.
What settles a CPI market?
The series and release the contract names (headline or core, year-over-year or month-over-month, seasonally adjusted or not) and the vintage of that figure. Contracts referencing "CPI" can therefore settle on different numbers drawn from the same publication.
Do data revisions affect these markets?
They can. Payroll estimates are revised twice in the two months after first publication and again at the annual benchmark, and seasonally adjusted CPI is revised on a rolling five-year window each February. Whether a revision affects a given contract depends on the vintage, the series and the settlement date it names.
What happens if a release is delayed?
The contract's own fallback terms govern. An appropriations lapse can suspend federal statistical publication, which is why data contracts generally specify what happens if the named release doesn't appear on schedule, though the specific provision varies by contract.