Easy money on Polymarket and Kalshi is disappearing as prop firms deploy AI agents
On July 28 and 29, the Federal Reserve meets to set its subsequent charge choice, and merchants will attempt to value the result by way of bonds, currencies, crypto and contracts that settle immediately on what the central financial institution declares.
Reuters polled 104 economists on July 21 and discovered each one in all them anticipated the Fed to carry at 3.50% to three.75%.
Kalshi’s July contract places 87% on that outcome, with roughly $29.7 million in quantity on the web page, and somebody nonetheless has to place a value on the opposite 13%.
That counterparty now consists of market makers, quantitative firms, funded-trading retailers and AI agents that watch costs, evaluate associated contracts and replace possibilities across the clock.
Institutions are testing occasion contracts, and brokers are wiring in liquidity suppliers. Funded-trading firms are beginning to deal with resolved contracts as a solution to determine merchants, human or algorithmic, who can value uncertainty higher than the group.
Together, these forces might deepen order books, pace up value discovery, and focus the sting amongst firms with the quickest infrastructure.
Combined month-to-month quantity throughout Kalshi and Polymarket reached its peak at $13.7 billion in June, with July already registering over $11 billion. These numbers present prediction markets already buying and selling at skilled scale.
Kalshi stated its annualized quantity greater than tripled over six months to $178 billion, that institutional quantity climbed 800%, and that it accomplished its first custom-made block commerce.
Clear Street, Marex and Jump Trading have every constructed a bit of the entry layer round that progress: Clear Street connects institutional purchasers to Kalshi, Marex works throughout each Kalshi and Polymarket infrastructure, and Jump helps establishments attain occasion markets immediately.
AQR, Susquehanna, and OKX have marketed specialist prediction-market roles on high of that build-out.
Corporate treasuries are testing these identical contracts to hedge tariff and regulatory exposure, a requirement that solely works if another person commits to pricing the opposite aspect of the commerce, constantly and at measurement.
Building a functioning market requires a provide aspect keen to cite each instructions, evaluate associated contracts throughout venues and correct a price the second it seems to be incorrect.
Measuring the sting
Louis Régis, founding father of the on-chain prop agency Propr and a former quantitative dealer at Credit Suisse, argued that occasion contracts make dealer choice extra rigorous than typical markets do, as the ability they reward is legible and the chance is bounded.
A contract resolves towards an outlined end result, so an allocator can look at whether or not a dealer constantly priced chance higher than the market. That take a look at isolates ability extra cleanly than a directional profit-and-loss report, the place market path and margin mix into the quantity.
The Foresight Arena benchmark estimates that detecting an actual fringe of two share factors with affordable statistical confidence takes about 350 resolved binary predictions, and confirming a one-point edge takes roughly 4 occasions as many.
A brief profitable streak on a handful of Fed or election contracts can nonetheless come from a favorable market pick, a correlated place, or a uncommon end result that occurred to land proper.
| What a funded agency can measure | Why it issues | Caveat |
|---|---|---|
| Probability calibration | Did the dealer repeatedly purchase possibilities that resolved too low or promote possibilities that resolved too high? | Needs many resolved contracts to separate ability from luck. |
| Performance after charges and slippage | Shows whether or not the sting survives actual execution prices. | Thin books could make paper edge disappear. |
| Drawdown management | Tests whether or not the dealer can survive dangerous occasion clusters. | Bounded draw back doesn’t remove correlated losses. |
| Market specialization | Reveals whether or not edge comes from macro, politics, crypto, sports activities or regulatory occasions. | Niche experience might not switch throughout classes. |
| Live-capital conversion | Shows whether or not simulated indicators are sturdy sufficient to be A-booked. | Nominal funding can overstate precise venue liquidity. |
| Sample measurement | Foresight Arena suggests small edges want a whole bunch of resolved predictions to confirm. | A scorching streak throughout a couple of main occasions is not sufficient. |
Propr plans to increase its analysis mannequin to Polymarket, letting merchants and AI agents qualify for accounts as much as $100,000 and maintain as a lot as $300,000 throughout a number of accounts, with an 80% revenue share as soon as they cross.
The agency treats each commerce as a sign, copying some onto the stay venue as A-booked positions and simulating the remaining internally as B-booked ones, crediting the dealer with the similar revenue and loss both approach.
Right now, Propr copies roughly 5% of its indicators onto a stay venue. The relaxation keep B-booked, a holding sample Régis attributes to gathering sufficient information to deploy treasury capital responsibly, and payouts settle on-chain in USDC no matter reserving methodology.
The execution downside
Régis expects AI agents to suit prediction markets particularly effectively: every contract follows a hard and fast construction, produces an observable value, and resolves towards a set rule.
An agent can watch a market and reprice it constantly, minute by minute, and Régis argued {that a} structured atmosphere plus fixed repricing provides as much as a genuine trading edge.
The Prediction Arena benchmark gave six frontier fashions $10,000 every and allow them to commerce autonomously on Kalshi and Polymarket between Jan. 12 and Mar. 9.
The fashions misplaced between 16% and 30.8% of their capital on Kalshi, and they averaged a smaller, still-negative 1.1% return on Polymarket. A separate working paper on changing forecasts into revenue argues that predictive accuracy solely turns into anticipated revenue underneath a proper betting strategy and sufficient liquidity to execute it.
Prediction markets could also be an unusually clear laboratory for AI merchants to check how good the fashions are at turning forecasts into worthwhile trades.
The vary of outcomes
In the bull case, funded merchants, market makers and agents provide sufficient stay capital to tighten spreads, deepen order books and pull Kalshi and Polymarket costs nearer collectively.
A January 2026 working paper examined frequent contracts throughout Polymarket, Kalshi, PredictIt and Robinhood. It discovered that Polymarket usually led Kalshi in value discovery when liquidity and buying and selling exercise ran increased, and that giant directional order circulate helped determine which venue moved first.
More stay capital might prolong that lead throughout extra contracts and compress the hole between platforms.
In the bear case, the sting concentrates in a handful of firms with the quickest infrastructure. Casual merchants lose constantly to better-informed counterparties, and liquidity thins out when an occasion will get hardest to cost.
Régis stated he is “assured concerning the path, not the magnitude.” One funded agency, even a rapidly increasing one, is nonetheless a small supply of circulate subsequent to a market that already strikes tens of billions a month.
| Scenario | What occurs | Who advantages | Main threat |
|---|---|---|---|
| Bull case: skilled liquidity flywheel | Funded merchants, market makers and AI agents tighten spreads, deepen books and appropriate stale costs sooner. | Institutions, venues, expert merchants and customers in search of higher costs. | Edge nonetheless concentrates, however market high quality improves sufficient to justify it. |
| Base case: professionalization stays selective | The deepest contracts round Fed choices, elections, sports activities and crypto get higher liquidity, whereas long-tail markets stay skinny. | Specialist merchants and venues with the strongest flagship markets. | Most prediction markets stay too shallow for institutional use. |
| Bear case: opposed choice dominates | Faster firms seize mispricings earlier than informal merchants can react, and liquidity disappears throughout hard-to-price moments. | Quant desks, bots and prop firms with superior infrastructure. | Prediction markets change into much less like crowd knowledge and extra like retail circulate buying and selling towards professionals. |
| Black swan: belief or authorized shock | A disputed decision, manipulation episode, or regulatory block slows institutional adoption. | Competitors with stronger compliance or settlement credibility. | Professional capital waits on the sidelines till market guidelines harden. |
The Fed’s hold on July 28 and 29 is already near settled earlier than the assertion even prints. The contest performs out within the remaining tail, the chance band sitting exterior that very same consensus, and within the moments as soon as CPI, GDP and payroll information power each contract to reprice.
The Bureau of Economic Analysis publishes its advance GDP estimate July 30, and the July employment report arrives Aug. 7. Each launch runs the identical competitors once more: whoever costs the shock first, or fixes a stale line quickest, retains the circulate.
Pricing these releases accurately, once more and once more, is what turns a dealer or a mannequin into one thing a funded-trading agency desires to again with actual capital.
The put up Easy money on Polymarket and Kalshi is disappearing as prop firms deploy AI agents appeared first on CryptoSlate.
