CPI Prediction Market Contracts: How They're Structured and Priced

TL;DR

• CPI, jobs report, and GDP contracts are structured as a ladder of range buckets (e.g., "0.2% to 0.3%"), not single yes-or-no bets, because the underlying number is continuous rather than binary.
• Each bucket trades as its own Yes/No contract. Add every bucket's price together for one release and the total should land close to $1.00 before fees, since exactly one bucket has to be correct.
• Most trading volume on these contracts happens in the 48 hours before release, because the calendar date is fixed and public months in advance.
• Kalshi, Polymarket, Limitless, Predict.Fun, and Opinion all list economic data contracts, but settlement source and contract format vary by platform.

Ever pulled up a CPI contract and wondered why there isn't just one bet for "inflation beats expectations"? You're not missing something. That contract doesn't exist. What you're trading instead is a narrow slice of next month's number, one contract for each range it could land in.

A CPI prediction market contract prices a narrow range of next month's inflation figure, not a single up-or-down call. Platforms split the release into a ladder of buckets, each trading as its own yes-or-no contract, because a single binary can't capture a continuous number the way it captures an election or a match result.

That structural difference is the whole story. Once you understand why CPI, jobs report, and GDP contracts get built as ranges instead of single bets on an outcome, the rest of the mechanics fall into place.

Why Do These Contracts Cluster Around Release Dates?

The Bureau of Labor Statistics publishes the Consumer Price Index on a fixed monthly schedule, announced a year in advance, always at 8:30am ET. The jobs report, officially the Employment Situation Summary, comes out the same way, typically the first Friday of the month. GDP arrives quarterly from the Bureau of Economic Analysis, with an advance estimate followed by two rounds of revisions.

Because the calendar is public and fixed, trading activity has a predictable shape. It builds in the days before a release as forecasters position around the consensus estimate, spikes hard in the minutes after the number prints, and then the contract resolves almost immediately. Compare that to an election contract, which can trade for a year with no single moment of resolution, or a World Cup winner contract that stays open for the length of a tournament. CPI and jobs contracts live and die in a single morning.

This is also why liquidity on economic data contracts looks different from liquidity on political or sports contracts. A presidential election market accumulates volume gradually over months. A CPI bucket contract can see most of its lifetime volume in the 48 hours before the release, because that is the only window where new information (bank forecasts, private payroll data, PMI prints) is still arriving.

How Do Buckets Get Built for a Continuous Number Like CPI?

Inflation doesn't happen or not happen. It lands somewhere on a number line, which means a single yes-or-no contract can't price it. Platforms solve this by chopping the possible outcomes into adjacent ranges and listing each range as its own binary contract.

Say the consensus forecast for monthly CPI is 0.3%. A platform might list contracts for "0.0% to 0.1%," "0.1% to 0.2%," "0.2% to 0.3%," "0.3% to 0.4%," and so on, each one a separate Yes/No market. If the bucket containing 0.3% trades at $0.35, that's a 35% implied probability the actual print lands in that specific tenth-of-a-point range. Add up every bucket's price and, before fees, the total lands close to $1.00, because exactly one bucket has to be correct. That's the same overround math behind every prediction market's book, just spread across many buckets instead of two.

This is the same yes-or-no mechanic that underlies every prediction market contract. A Yes contract pays $1 if the outcome falls in its range, and $0 if it doesn't. What changes for economic data is that you're not pricing one proposition, you're pricing an entire distribution, one bucket at a time.

The bucket that sits closest to the consensus forecast usually carries the highest price and the deepest liquidity, since that's where most forecasters agree the number will land. The tail buckets, a much hotter or much cooler print than expected, trade thin and cheap, which is exactly where the interesting risk sits.

Jobs report contracts work the same way. Instead of a single "will nonfarm payrolls beat expectations" contract, platforms list a ladder of buckets across a range like 100,000 to 150,000 jobs, 150,000 to 200,000, and so on.

GDP contracts do the same thing around the advance estimate, and this is where the revision problem shows up. The Bureau of Economic Analysis issues an advance estimate, then revises it twice over the following two months, and BLS payroll figures get revised in the two subsequent jobs reports as well.

A well-built contract states plainly which release it settles on, first print or a specific revision, because those numbers can move enough to flip a bucket after the fact.

What Makes Economic Data Contracts Different From a Single-Event Contract?

A political contract like "will this candidate win the election" or a sports contract like "will this team win the tournament" resolves on one discrete fact. There's a winner and a loser, full stop. Economic data contracts resolve against a number that has to be sorted into a range, which means the market design itself is doing more work before a single position is ever taken.

It also means economic data contracts are built as recurring contracts in a way election and tournament contracts aren't. The same CPI bucket structure reappears every month, which lets you watch how the consensus forecast shifts release over release, something you can't do with a one-off event. That recurring cadence is what makes this corner of prediction markets closer to a running data feed than a single wager on an outcome.

Where Do the Five Platforms Differ on Economic Data Contracts?

PredictionHero tracks CPI, jobs report, and GDP contracts across all five platforms it aggregates, and the structural differences between them matter more here than on political or sports markets, because settlement source and contract format vary.

Kalshi is worth starting with because it was built as a CFTC-regulated exchange from day one, which means its CPI and jobs report contracts settle directly against the government release with a retail-friendly, bank-funded account. That regulatory footing is also why its economic data contracts tend to carry the deepest US-based liquidity around a release.

Polymarket runs the largest prediction market by volume globally, and its economic data contracts draw a more international, crypto-native base. Positions are collateralized in crypto and resolved through an oracle rather than a direct exchange feed, which opens the contracts to traders outside the US that a bank-account-only platform can't reach.

Opinion is the platform built specifically around this category. Sports and political contracts sit alongside its core focus rather than leading it. FOMC decisions, CPI prints, and GDP data are what its product was designed for first, which shows up in how many buckets it lists and how tightly they're spaced around the consensus forecast.

Limitless runs on-chain with its own token and tends to specialize in shorter-duration contracts, which fits the fast resolve-and-move-on nature of a monthly data print. Predict.Fun adds a mechanic none of the others do: collateral sitting in an open position earns yield while it waits for the release, which matters for traders holding a bucket position through several days of pre-release positioning.

PredictionHero aggregates publicly available prediction market data for informational purposes only. This is not financial advice. Prediction markets may not be available in all jurisdictions.