Prop trading firms, market makers and AI agents are reshaping liquidity on Polymarket and Kalshi, turning what was once a retail-friendly arbitrage into an infrastructure race. Combined monthly volume across the two venues peaked at $13.7B in June and is already past $11B in July, with Kalshi's annualized volume tripling over six months to $178B. Kalshi says institutional volume climbed 800% and the venue just executed its first customized block trade, a signal that liquidity providers are now committing size on both sides of the book.
The access layer is filling out fast: Clear Street wires institutional clients to Kalshi, Marex works across both venues' infrastructure, and Jump Trading helps desks reach event markets directly. AQR, Susquehanna and OKX have advertised specialist prediction-market roles on top of that build-out, and corporate treasuries are testing the same contracts to hedge tariff and regulatory exposure.
Why it matters
Edge on event contracts rewards a skill that is unusually legible: pricing probability better than the crowd on a defined resolution. Louis Régis, founder of on-chain prop firm Propr and a former Credit Suisse quant, argues that bounded-risk, rule-resolved outcomes let allocators isolate trader skill more cleanly than a directional P&L record does. The Foresight Arena benchmark estimates that confirming a genuine two-point edge takes around 350 resolved binary predictions, and a one-point edge roughly four times that, which means short winning streaks on a handful of Fed or election contracts can still come from correlated positions or lucky event picks.
AI agents fit the format naturally. Each contract follows a fixed structure, produces an observable price, and resolves against a set rule, so an agent can reprice continuously, minute by minute. The Prediction Arena benchmark gave six frontier models $10K each to trade autonomously on Kalshi and Polymarket between January 12 and March 9; the models lost between 16% and 30.8% on Kalshi and averaged a still-negative 1.1% return on Polymarket, a reminder that forecasting accuracy does not automatically convert to expected profit without a proper betting strategy and enough liquidity to execute.
Market impact
Propr plans to extend its evaluation model to Polymarket, letting traders and AI agents qualify for accounts up to $100K and hold as much as $300K across multiple accounts with an 80% profit share.
Frequently asked questions
-
Why is edge on Polymarket and Kalshi shrinking for casual traders?
Prop trading firms, quantitative shops and AI agents are now quoting both sides of event-contract order books around the clock. Their infrastructure reprices faster than retail can react, so the mispricings that once paid casual traders are getting closed before they can be traded.
-
How much volume are prediction markets doing now?
Combined monthly volume across Kalshi and Polymarket peaked at $13.7B in June, with July already past $11B. Kalshi's annualized volume more than tripled over six months to $178B, and the venue says institutional volume climbed 800% and completed its first customized block trade.
-
What is Propr and how does its evaluation model work?
Propr is an on-chain prop firm founded by former Credit Suisse quant Louis Régis. It evaluates traders and AI agents on event contracts, plans to extend to Polymarket with accounts up to $100K and up to $300K across multiple accounts, and offers an 80% profit share. About 5% of its signals are A-booked onto a live…
-
How well did AI agents actually perform on prediction markets?
The Prediction Arena benchmark gave six frontier models $10K each to trade autonomously on Kalshi and Polymarket between January 12 and March 9. The models lost between 16% and 30.8% on Kalshi and averaged a still-negative 1.1% return on Polymarket, showing that forecasting accuracy alone does not convert to expected…
-
Why does the Fed meeting matter for prediction-market liquidity right now?
The Fed's July 28 to 29 decision is already priced at 87% for a hold at 3.50% to 3.75% on Kalshi, with about $29.7M in volume on that contract. The real competition is in the remaining 13% tail and in the repricing that follows July 30's advance GDP and the August 7 employment report, where whoever prices the surprise…
CryptoSlate