A prediction market is a venue where the price of a contract is the probability of an event. If a contract that pays $1 if the Federal Reserve cuts rates in July is trading at $0.42, the market is saying there is a 42% chance of a cut. That is the entire idea. Everything else — the order books, the settlement contracts, the regulatory overlays, the hundreds of millions of dollars now flowing through these venues each week — is just plumbing around that single, very old idea that crowds aggregating money on a question tend to be more honest than crowds aggregating opinions.
Most people who follow finance still cannot define a prediction market in one sentence. That gap is the opportunity. The asset class doubled again in the last twelve months, regulators in three jurisdictions moved from hostility to active framework-building, and two platforms crossed valuations that a decade of academic interest never produced. This is the post we wished existed when we started building the infrastructure layer underneath it.
What a prediction market actually is
A prediction market is a contract market for the outcome of a real-world event. The simplest form is binary: a contract pays $1 if an event happens, $0 if it doesn't. Traders buy and sell those contracts during the life of the market, and the price at any moment is the market's implied probability of the event.
Three primitives make up every prediction market:
- The contract. A clearly specified event with a clearly specified resolution date. "Will the U.S. CPI year-over-year reading be above 3.0% on the September 2026 release?" is a contract. "Will inflation stay high?" is not — too vague to settle.
- The order book. A live two-sided market where traders post bids ("I'll pay $0.42 for YES") and offers ("I'll sell YES at $0.45"). The midpoint is the implied probability. The spread is the cost of immediate execution.
- The settlement rule. A pre-agreed mechanism that determines, on the resolution date, what actually happened — and pays out $1 to one side and $0 to the other.
The cleanest mental model is that a prediction market is a tiny piece of derivative-style infrastructure with a clearly bounded payoff. You are not "betting" in the casino sense, where the house has an edge. You are buying a contract whose terminal value is determined by an external fact, from another trader who disagrees with you about the probability.
How the order book and pricing work
Almost every modern prediction market uses a Central Limit Order Book (CLOB). It works the same way as a stock exchange's order book, with a few constraints specific to prediction contracts.
Suppose a contract on "Fed cuts rates in July" is trading. A few resting orders sit in the book:
- BID: $0.41 for 200 contracts
- BID: $0.40 for 1,000 contracts
- ASK: $0.43 for 500 contracts
- ASK: $0.45 for 2,000 contracts
If a new trader believes the cut is a near-certainty, she might lift the $0.43 ask, paying 200 × $0.43 = $86 to receive 200 YES contracts. If she's right, she gets 200 × $1 = $200 at settlement, a 132% return on the cash she put up. If she's wrong, those contracts settle at $0 and she loses the $86.
The other side, the seller of those YES contracts, is short the event. Her view is that the cut is less likely than 43%. She collects $86 upfront and is on the hook for $200 if the cut happens. Either she holds to settlement, or she buys back the position later at a different price.
There are two important details that make prediction markets different from a traditional derivative book:
- Capped collateral. Because YES + NO contracts always sum to $1 by construction, the maximum loss for either side is bounded. That lets the system run without margin calls — the worst case is fully collateralised at order entry.
- No funding rate, no perpetual rollover. Each market has a hard resolution date. The implied probability isn't a rolling guess about the future; it converges to 0 or 1 at a known time.
This is why prediction-market liquidity feels different from spot-crypto liquidity. The book is denser near the implied probability, spreads tighten as resolution approaches, and the entire structure behaves more like a short-dated option than a perpetual swap.
How outcomes get resolved
The most important part of any prediction market is the part that happens once: how it determines what actually occurred. If resolution is opaque or political, the entire market loses credibility. If resolution is robust, the platform can host thousands of markets without the operator becoming a single point of trust.
There are three resolution models in production today:
- Centralised (regulated exchange). Kalshi, the CFTC-regulated US venue, resolves contracts through its internal market integrity team using pre-published data sources. This is how most traditional derivatives markets resolve.
- Optimistic oracle (UMA). Polymarket and most on-chain markets use UMA's Optimistic Oracle. A proposer submits the outcome with a bond. If nobody disputes within a challenge window, that outcome is final. If somebody disputes (also bonded), UMA token holders vote on the truth.
- Aggregated data feeds (Chainlink, custom). For markets settling on numerical data — sports scores, asset prices, weather — many venues use a deterministic feed plus a tie-breaking dispute window.
Each model trades off speed, cost, and decentralisation differently. We've written a deep-dive on this — see how resolution actually works under the hood for the full mechanics.
A short history: from Iowa to Polymarket
Prediction markets are not new. They're an old idea that finally found the right substrate.
- 1988 — Iowa Electronic Markets. A research project at the University of Iowa let academics and a few hundred traders bet small amounts on US elections. The IEM consistently out-predicted polling for two decades. It was operational proof that markets aggregate information.
- 2003–2013 — InTrade. The first popular consumer prediction market, based in Ireland. It hit hundreds of millions in annual volume before the CFTC went after it. InTrade shut down in 2013, ostensibly over an internal accounting issue but really because the US regulatory environment didn't tolerate offshore venues.
- 2016 — Augur. The first on-chain prediction market, built on Ethereum. Augur showed that the resolution problem could be solved decentrally. It also showed the UX wasn't ready: markets settled weeks late, gas costs ate edge, and most users struggled with on-chain wallets.
- 2020 — Polymarket. The first mass-market on-chain venue. Polymarket adopted USDC settlement on Polygon and adapted UMA's optimistic oracle for resolution. The 2024 US election cycle pushed it past $9B in cumulative volume.
- 2021 — Kalshi. The first CFTC-regulated event contract exchange in the US. Kalshi's regulated status is what unlocked institutional flow — which is what pushed the category over $18B/month combined.
The pattern is consistent across every prior wave: the technology is fine, the user demand is real, and the bottleneck is always infrastructure plus regulation. Both of those are now resolved.
The current state of the market
What changed in the last twenty-four months is not the idea. It's that two stable venues now exist (one regulated, one decentralised), they have proven they can carry billions of dollars of monthly volume without breaking, and a generation of traders who came up trading crypto perpetuals find prediction-market mechanics intuitive.
| Metric | End of 2023 | End of 2024 | Q1 2026 |
|---|---|---|---|
| Combined monthly volume | $340M | $3.1B | $18.3B |
| Polymarket monthly visits | 1.8M | 12M | 32M |
| Kalshi annual run rate | — | $420M | $1.4B |
| Active markets / week | ≈ 600 | ≈ 2,400 | ≈ 9,800 |
| Institutional volume share | <3% | ≈ 14% | ≈ 31% |
The institutional share is the most important line in that table. The whole shape of an asset class changes when professional desks start quoting it. Spreads compress, position sizes scale, and the cost of executing a $5M view moves from "impossible" to "routine."
The category has crossed the threshold where it can no longer be dismissed as a niche. Three of the largest broker-dealers we speak with have a prediction-market initiative on their 2026 roadmap.
Why 95% of the world hasn't accessed them yet
Despite that growth, prediction markets are still functionally unavailable to most of the people who'd trade them. The gap is a combination of geography and brand.
Geography. Polymarket is geo-blocked in the US (regulatory). Kalshi is unavailable outside the US (also regulatory, mirrored). Together, that means a trader in São Paulo, London, or Singapore who wants to take a view on US monetary policy has to either route through a crypto-native interface that wasn't built for them or skip the trade. Roughly two-thirds of Polymarket's traffic already comes from outside the US, on a platform that doesn't market itself there.
Brand. A retail trader in Brazil already trusts a local brokerage with their savings. They don't trust a foreign-language website without a regulated counterpart in their jurisdiction. The local brokerage has the relationship; it just doesn't have the infrastructure to host the product.
This is the gap Kuest exists to close. The protocol thesis is simple: the local brokerage shouldn't have to build prediction-market infrastructure to offer prediction markets. They should be able to license it, white-label it, and earn a fee on every trade — the same way merchants run on Shopify without writing payment-processing code.
How traders actually use prediction markets
The textbook answer is that traders are expressing views on event probabilities. The actual answer is more interesting and gives you a better feel for why volume keeps growing.
Three trader profiles dominate the order books today.
The macro hedger. This is the professional desk we mentioned earlier. A fund with a duration view holds a CPI contract not as a view on inflation per se but as a hedge against a specific scenario in their main book. A 0.30 contract on "CPI prints above 3.2%" becomes a $300k-notional hedge against a $20M Treasury position. The asymmetry is what makes it useful — the worst case is that the hedge expires worthless, not that the desk takes margin damage.
The narrative trader. This is the retail volume that used to live in crypto perpetuals and is migrating to prediction markets because the unit economics are better. A view on "the Fed cuts in July" is expressible in a perp through SOFR-curve approximations, but you carry funding, you carry overnight rolls, and you carry execution risk. A binary contract removes all of that. You pay your $0.42, you hold to settlement, you get $1 or $0. The product is honest in a way perps aren't.
The arbitrageur. This is the under-discussed profile. There's a constant stream of arbitrage between Polymarket, Kalshi, and the implied probabilities embedded in traditional derivatives. When SOFR-options-implied Fed cut probability is 47% and Kalshi is trading at 41%, somebody pulls those into line. As the category matures, the arbitrage corridors narrow and the cross-venue correlation tightens. This is what makes the asset class behave increasingly like a mainstream derivative product.
What you don't see, in any meaningful size yet, is the institutional allocator using prediction markets for portfolio-level risk expression. That's the next leg of the curve. When a $40B multi-strategy fund books a $50M sleeve in macro prediction-market exposure as a directional view, the venue infrastructure has to be ready to absorb size at that scale. The platforms that will absorb that flow are being built now.
What's coming next
Three forces are converging through 2026 and 2027 that we expect to push the category toward $50B+ in monthly volume:
- Regulatory clarity in the EU and Brazil. The CFTC's no-action posture toward Kalshi is being mirrored by ESMA discussion drafts and by Brazilian Ministry of Finance consultations. Once domestic regulated event-contract exchanges become possible in those jurisdictions, the geographic ceiling lifts.
- Institutional flow becomes a majority. Once professional desks account for >50% of volume on Kalshi and Polymarket, a quote stack forms that absorbs much larger orders. The category starts behaving like a real asset class — option-style risk sleeves at scale.
- The infrastructure layer matures. White-label deploys, shared liquidity, and managed-resolution rails compress the time from "operator decides to launch" to "platform live with real volume" from quarters to days. That changes who can be an operator. Today it's a venue-builder. Soon it's any brokerage, media company, or creator with an audience.
If you're researching this category as a potential operator, distributor, or institutional allocator, the next 18 months are the window in which the architecture will be set. The platforms that won the perpetuals wave were the ones that committed early; the same will be true here.