Prediction markets are booming. The investable proxies are not.

The question arriving in portfolio reviews is usually one sentence: how do we own prediction markets? The answer an advisor can defend is that the two venues doing the volume cannot be bought, and the tokens sold as the closest substitutes are paid from a different revenue line.

A contract pays a dollar if a future event happens and nothing if it does not, so a price between zero and one reads as the market's probability. Citywire RIA reports that combined trading on Kalshi and Polymarket, described as the two largest platforms, climbed from $4.51 billion in September 2025 to $48.9 billion in July 2026, roughly eleven times the volume in ten months. Intercontinental Exchange, which owns the New York Stock Exchange, has put about $1.6 billion into Polymarket and distributes its data to clients. Robinhood's event contracts produced $156 million in the second quarter of 2026, more than its equities trading brought in.

The two platforms are not purchasable. Kalshi is private, valued at $22 billion, and its chief executive said in June it would not list this year, with reports pointing to 2027 or later; Polymarket is private and has said it intends to issue a token at some point, without giving a date. What remains is infrastructure: Hyperliquid, which runs its own prediction market and charges a fee on the trade, and Solana, which does not run one but collects a general transaction fee from the apps sitting on it. Their tokens are HYPE and SOL.

Citywire's figures argue against treating either token as a claim on the activity. Hyperliquid's prediction market contributed $3,830 of revenue in the first 27 days of August, and its volumes have fallen every month since June. Solana does not run a prediction market at all, and in the largest case it earns nothing, because Polymarket's contracts run on Polygon. Polygon sharpens the point: it hosts the largest prediction market in the world, Polymarket traded $7.7 billion in January and $10.6 billion in March, and over those same two months Polygon's network revenue fell from $3.37 million to $976,000 and has stayed below its January level since. The cause was not thin activity: Polygon changed its fee mechanism in March 2026 to run the network at full capacity, which made transactions cheaper and more predictable.

The portfolio conversation runs into a deeper problem than binary-payoff volatility: the proxy is decoupled. The biggest venue's volume rose about 38% between January and March while the network hosting it lost roughly 71% of its revenue. An advisor buying HYPE or SOL to give a client event-contract exposure has bought a crypto network's fee schedule and relabeled it a theme. The sizing math for a position that only pays out if one of those networks captures volume the leaders route elsewhere belongs in a speculative sleeve, disclosed as such. The dates that would change the answer are a Polymarket token or a Kalshi listing. Neither has one.