Prediction markets are supply-constrained.

The dominant story about prediction markets in 2026 is liquidity. Why are spreads so wide? Why does a $5,000 buy move the price three cents? The answer everyone reaches for is "we need more capital." It is the wrong answer. The constraint is not capital. It is the number of contracts that exist.

The episodic-event problem

A US presidential election produces about twenty markets of meaningful liquidity. The Super Bowl produces a few dozen prop bets. A typical week of NFL produces sixteen game outcomes. Sum it across a calendar year and the total number of high-liquidity events on the largest prediction-market platforms is in the low thousands.

That is the entire supply side. Every dollar of platform GMV in 2026 is allocated across this fixed pool. The math of liquidity-per-contract is the math of capital divided by contract count. With the contract count constrained, increasing capital just makes the existing contracts deeper — a useful improvement, but it does not change the shape of the platform.

This is the supply-constrained regime. The question is whether the constraint is fundamental or whether it is removable.

Continuous flow as a missing primitive

The reason elections and sports anchor prediction markets is they are decisively-resolvable real-world events. The platform's machinery (orderbook, matching, settlement) is well-suited to those properties. The fact that decisively-resolvable events are episodic was historically an accidental feature of the categories the platforms started with, not an architectural requirement.

What changes if a settleable category is continuously flowing rather than episodic? Three things, simultaneously:

First, total contract count grows by orders of magnitude. The 2024 US election produced roughly twenty contracts of meaningful liquidity over a year. Aviation produces 500,000 contracts per day, every day, for as long as commercial flight exists. The category-level supply is not 25× larger; it is 9,000× larger.

Second, the platform's economics change. Listing-and-resolution overhead amortizes across orders of magnitude more contracts. Marketing cost per contract drops to near-zero because individual contracts no longer need promotion — the category does. Capital that previously concentrated on twelve weekly events now spreads across thousands of daily ones.

Third, the user behavior pattern changes. Election-cycle prediction markets have spiky engagement: users open the platform when there is news. Continuous-flow categories invite a different pattern — checking flight-day positions, hedging an upcoming travel itinerary, settling micro-stakes hourly. This is the engagement pattern of sportsbooks during a season, applied to a category that has no off-season.

The hidden ceiling on platform GMV

If you charted the GMV of the largest prediction-market platforms over the last five years, you would see a bursty, election-correlated curve. 2020 spike, 2024 spike, lower troughs in between. The peak is what gets reported; the trough is what gets ignored. Both are determined by the same factor: how many decisively-resolvable events are happening this month.

Capital does not solve this. Sophisticated traders do not solve this. UX improvements do not solve this. Market makers do not solve this. The single variable that moves the curve is contract count, and contract count is gated by the categories the platform can settle.

Why oracles, not exchanges, are the moat in 2026

For most of prediction-market history, the moat was the platform itself: matching engine, custody, regulatory compliance, brand. These remain real and important. But they are no longer the binding constraint on growth.

The binding constraint is the data layer. A platform without a settlement-grade aviation oracle cannot list aviation contracts, full stop. The moat shifts from "who has the better matching engine" to "who has the deeper oracle category coverage."

This is the same thesis that played out in DeFi between 2020 and 2025. Early DEX competition was about AMM curve design and routing; the lasting value accrued to oracles (Chainlink, Pyth) because they were the rate-limiting input to every settled financial primitive. Prediction markets are running through the same evolution, two years behind.

What aviation specifically changes

Aviation is the first continuous-flow category that meets the bar a prediction-market platform requires: deterministic resolution, observable settlement, machine-readable signal. It is not the only such category — weather, ride-share supply, energy demand are all candidate continuous-flow oracles — but it is the first one that is both technically ready and commercially relevant at scale.

For a platform like Polymarket, plugging into an aviation oracle adds 500,000 contracts per day without changing matching logic, custody, or capital. For a platform like Kalshi, the deterministic-resolution bar is already met. For sportsbooks, the integration is a new contract type on top of existing infrastructure. The cost is a webhook subscription. The yield is the entire upside of removing the supply constraint.

Why now

The reason this hasn't already happened: oracle infrastructure for aviation didn't exist at settlement-grade reliability until roughly 2024. Real-time aircraft telemetry coverage at the necessary density, automated outcome attestation, and the historical archive depth needed for detection models all reached production-readiness in roughly the same eighteen-month window. Before that, the math was the same; the integration was a custom build that no platform had a budget for.

The category infrastructure is now production-ready. The supply constraint can be removed. The platforms that integrate first capture the upside of being the first to offer the category at scale.