Retail Dividend Fades, Predicting an AI Arms Race in the Market

On the night of the Federal Reserve's July meeting, a "quantitative shadow war" over pricing power.

Written by: Gino Matos

Compiled by: Saoirse, Foresight News

From July 28 to 29, the Federal Reserve will hold a meeting to finalize a new round of interest rate decisions. Traders will use bonds, foreign exchange, cryptocurrencies, and event contracts directly anchored to central bank decisions to speculate on price movements resulting from the decisions.

On July 21, Reuters conducted a survey of 104 economists, all of whom expect the Federal Reserve to maintain the interest rate range at 3.50%–3.75%. The probability of this outcome in Kalshi's July contracts reached 87%, with a transaction volume of approximately $29.7 million, while the market still needs counterparties to quote for the remaining 13% possibility.

Currently, these counterparties include market makers, quantitative institutions, asset management proprietary teams, and AI agents. These programs monitor prices around the clock, compare similar contracts across platforms, and continuously update the probabilities of events occurring.

Various institutions are testing event contracts, and brokers are continuously introducing liquidity providers. Asset management proprietary firms have begun to use settled contracts as a benchmark for filtering traders—whether the trader is human or algorithmic, those who can price uncertainty more accurately than the market at large will be given special attention.

The convergence of multiple forces can thicken the order book and accelerate price discovery, but trading advantages will also concentrate among the institutions with the fastest infrastructure.

The total monthly trading volume of the two major platforms, Kalshi and Polymarket, reached a peak of $13.7 billion in June, and July's trading volume has already surpassed $11 billion. Data shows that the trading scale of the prediction market has reached a level comparable to professional institutions.

The chart shows that Kalshi and Polymarket's monthly trading volume hit a high of $13.7 billion in June, with Kalshi's annualized trading volume reaching $178 billion.

Kalshi stated that its annualized trading volume has more than doubled in six months to $178 billion, with institutional trading volume increasing by as much as 800%, and the platform has completed its first customized block trade.

Clear Street, Marex, and Jump Trading are all building access channels in line with this growth trend: Clear Street helps institutional clients connect to Kalshi; Marex connects both Kalshi and Polymarket; Jump Trading assists institutions in directly participating in event market trading. Additionally, AQR, Susquehanna, and OKX have all released recruitment information for professional trading positions in the prediction market.

Corporate finance departments are also experimenting with these contracts to hedge against tariff risks and exposures brought about by regulatory policies. However, the premise for this hedging demand to be valid is that the market has counterparties capable of continuously and significantly taking on reverse position quotes.

To build a well-functioning market, suppliers need to be willing to quote both ways, compare similar contracts across platforms, and immediately correct pricing when significant deviations occur.

Measuring Trading Advantages

Louis Régis, founder of on-chain proprietary trading firm Propr and former quantitative trader at Credit Suisse, proposed that compared to traditional financial markets, the screening criteria for traders in event contracts are much stricter. These contracts can clearly reflect a trader's judgment ability, while the risk boundaries are clear and controllable.

A contract ultimately anchors to a clear result settlement, allowing capital providers to intuitively judge whether the trader can consistently provide probability pricing superior to market consensus. Relying on event contracts to identify trading ability is far purer than simply looking at directional trading profit and loss records—ordinary profit and loss can easily be affected by market trends, margin fluctuations, and other factors.

Foresight Arena's benchmark calculations show that to confirm a stable trading advantage of 2 percentage points with reasonable statistical confidence, approximately 350 completed binary prediction contracts are needed; to verify a 1 percentage point advantage, the required sample size is about four times that of the former.

Relying solely on a few contracts related to Federal Reserve decisions or elections to achieve short-term profits may simply be a matter of selecting favorable trading targets, coincidental position correlations, or merely betting on random events, and does not represent long-term capability.

Propr plans to expand this evaluation system to Polymarket. Traders and AI agents who pass the assessment can receive a maximum trading limit of $100,000 per account, with a total limit of $300,000 across multiple accounts, and a profit-sharing ratio of up to 80%.

The company views each trade as an effective signal, with part of the signals replicated to the live trading platform as A-book positions, while the rest run in the system internally, classified as B-book. Regardless of the method, traders will receive equal standards for profit and loss accounting.

Currently, Propr only deploys about 5% of trading signals to the real trading market, with the remaining signals kept in internal simulation. Louis Régis stated that this model is adopted to accumulate sufficient data and prudently allocate proprietary funds. Whether A-book or B-book, the profits are ultimately settled on-chain in USDC.

Challenges at the Execution Level

Louis Régis believes that prediction markets are naturally suited for AI agents: each contract structure is standardized, prices can be observed in real-time, and settlements are completed based on fixed rules.

Agents can continuously monitor the market and update pricing every minute. Louis Régis stated that the combination of a standardized market environment and uninterrupted re-pricing capability can theoretically create a solid trading advantage.

Prediction Arena conducts benchmark tests: six cutting-edge AI models are allocated $10,000 each to trade independently on Kalshi and Polymarket from January 12 to March 9. The results show that the models experienced losses ranging from 16% to 30.8% on Kalshi; on Polymarket, the average loss was smaller, but still recorded negative returns, with an average drawdown of 1.1%. Another research paper pointed out that to convert prediction accuracy into stable profits, a reasonable betting strategy must be paired with sufficient liquidity to support the strategy's implementation.

Prediction markets can serve as an excellent testing ground to evaluate whether AI trading models can translate predictive insights into profitable trades.

Future Market Evolution Prospects

In an optimistic scenario, traders, market makers, and AI agents with funding support will bring ample real trading capital, narrowing bid-ask spreads, thickening order books, and bringing the price levels of Kalshi and Polymarket closer together.

A research paper from January 2026 analyzed similar contracts on Polymarket, Kalshi, PredictIt, and Robinhood platforms. The study found that when liquidity and trading activity are high, Polymarket often dominates price discovery, and large-scale one-way order flows determine which platform adjusts prices first. More real trading capital entering the market is expected to further expand the leading advantages of top platforms across more contracts and narrow the price gaps between major platforms.

In a pessimistic scenario, trading advantages will concentrate in the hands of a few institutions with top-notch infrastructure. Ordinary retail traders will continue to lose to counterparties with better information; when the market struggles to price events, liquidity will quickly shrink.

Louis Régis stated: "I am confident about the direction of development, but I cannot predict the final scale." Even if a proprietary institution expands rapidly, compared to a market where monthly trading volumes have reached hundreds of billions of dollars, the trading volume that a single institution can provide remains very limited.

The market has already anticipated the expectation that the Federal Reserve will maintain interest rates at their current level from July 28 to 29; before the official announcement, mainstream expectations have basically been set. The real competition exists in the tail ranges, which is the probability range that deviates from this consensus; when CPI, GDP, and non-farm payroll data are released, all contracts will face a concentrated repricing window.

The U.S. Bureau of Economic Analysis will release the preliminary GDP estimate on July 30, and the July employment report will be published on August 7. Each round of data releases will stage the same competition: whoever can predict unexpected data first, or correct outdated pricing the fastest, will grasp the trading order flow.

The ability to continuously and accurately price such data trends is key for a trader or a model to gain support from proprietary trading institutions.

Disclaimer: This content is provided for general branding and informational purposes only and doesn't constitute financial, investment, legal, or tax advice. Any events, rewards, online events, or related information mentioned herein should not be considered a recommendation, solicitation, or invitation to purchase, sell, trade, or otherwise deal in any crypto assets or to use any services. Crypto assets are highly volatile and may result in loss. WEEX services and online events may not be available in all regions and are subject to applicable laws, regulations, and eligibility requirements. You are responsible for ensuring that your use of WEEX services complies with local laws and for carefully assessing the risks before participating in any crypto-related activities.