Skip to main content

The Paradox at the Heart of Prediction Markets: Kalshi and Polymarket Are Banning the Traders Who Make Them Work

· 12 min read
Dora Noda
Dora Noda
Software Engineer

In April 2026, the two biggest prediction markets on the planet did something their own theoretical foundations say they should not do: they started kicking out the smartest people in the room.

Kalshi and Polymarket — between them clearing more than $66 billion in year-to-date notional volume — rolled out coordinated bans on the trades they were arguably built to price. Politicians can no longer wager on their own campaigns. Athletes are blocked from trading in their own leagues. Employees are barred from event contracts tied to their employers. Kalshi has gone so far as to ship "preemptive technological guardrails" that block these users before an order ever reaches the book.

There is just one problem. Robin Hanson — the George Mason economist who is, more than anyone else, the intellectual father of modern prediction markets — has spent the last week on the record arguing that insiders are not a bug. They are the entire point.

Welcome to the strangest market microstructure debate of 2026.

What Actually Changed in April​

The triggering events happened in clusters. On April 22, Kalshi announced enforcement actions against three U.S. congressional candidates — Mark Moran (a former Virginia Democratic Senate primary candidate now running as an independent), Minnesota state Senator Matt Klein, and Texas Republican Ezekiel Enriquez — for "political insider trading" on their own races. Each was suspended from the platform for five years.

The penalties revealed something interesting about scale. Moran was hit with a $6,229.30 fine plus disgorgement of profits. Klein got $539.85. Enriquez got $784.20. The actual wagers? Klein and Enriquez each bet less than $100. Moran himself confirmed on social media he "traded $100 on myself," and added the kicker: he did it "because I wanted to get caught" — to test whether Kalshi would actually enforce a rule it had announced but never invoked.

Days later, both Kalshi and Polymarket published formal policies extending the prohibition far beyond politicians:

  • Candidates and their staff cannot trade on their own elections.
  • Pro athletes, coaches, and front-office personnel cannot trade markets tied to their leagues.
  • Employees cannot trade event contracts on outcomes tied to their employer.

Then came Washington. On April 24, Senator Bernie Moreno (R-OH) introduced a Senate resolution prohibiting senators from entering "any agreement, contract, or transaction" that provides a financial benefit based on event outcomes. That resolution sits alongside the End Prediction Market Corruption Act (S. 4017), introduced in March 2026 by Senator Jeff Merkley (D-OR) with Klobuchar, Van Hollen, Schiff, and Gillibrand, which would amend the Commodity Exchange Act to ban the President, Vice President, and all members of Congress from event contracts entirely — with civil penalties of up to $10,000 per violation and explicit authority for the CFTC to write rules against misuse of material nonpublic information.

In short, the regulatory floor and the platform ceiling are converging on the same answer: certain people should not be allowed to trade on certain outcomes. Ever.

Hanson's Counterpunch​

Robin Hanson did not take this quietly. In interviews tied to the April 26 Fortune feature, the economist who designed the market scoring rule that powers most modern prediction-market liquidity — and who created the futarchy framework, in which national policies would be selected by prediction markets themselves — argued that the new bans inverted the entire purpose of the technology.

His framing is brutal in its simplicity:

The point of prediction markets is to produce accurate prices, and insiders have the most accurate information. Banning them is like banning your best sources from contributing to a newspaper.

This is not a fringe position. It is the position. Hanson's foundational 2008 paper "Insider Trading and Prediction Markets" argues explicitly that insider trading in event contracts is welfare-improving, because the markets exist for price discovery, not to allocate scarce capital. There is no productive enterprise being deprived of investment when an insider trades a Polymarket contract on the next FOMC decision. There is only a probability estimate that becomes more accurate, faster, the moment that informed trade settles.

Hanson does grant the tradeoff. "There's organizations that want to keep secrets," he conceded, "and then there's a larger world that often wants to know those secrets." That is the tension prediction markets exist to resolve — by paying insiders to leak, in the form of profits, what they would otherwise be paid to suppress.

The platforms are now choosing the side of the secret-keepers. Hanson's argument is that they are betraying the wisdom-of-crowds thesis they marketed to retail and the price-discovery thesis they marketed to regulators.

The 3% Problem​

A peculiar piece of empirical evidence dropped into this debate the same week. CoinDesk reported on a 2026 academic study analyzing every Polymarket trade from 2023 through 2025. The headline finding: roughly 3% of traders drive most of the platform's accuracy. The "wisdom of the crowd" framing — that prediction markets work because thousands of small bettors aggregate their slivers of insight — is empirically wrong, at least at Polymarket's current scale.

What works is wisdom of the informed few.

Read in light of the new bans, this is awkward. If 3% of traders generate the price signal, and that 3% is disproportionately composed of people with privileged information, then aggressively filtering out exactly those traders is a direct attack on the mechanism that makes the prices worth quoting in the first place. Every politician barred from their own race is — on the Hanson view — a small loss of accuracy. Every employee barred from their employer's event contract is a degradation of what the market is meant to signal.

The platforms are betting the opposite: that they can survive on the noisier wisdom of the remaining 97%, plus institutional liquidity providers, plus the compounding benefit of being seen as legitimate and law-abiding.

The Horse Racing Precedent​

To see why the platforms are choosing integrity over information, look at the markets that already resolved this tension.

Horse racing, in nearly every U.S. state racing commission, prohibits owners, trainers, and jockeys from betting on their own races. The NCAA bans athletes from betting on any sport in which they compete; the NFL, NBA, and MLB ban players, coaches, and front-office staff from sportsbook activity tied to their leagues. None of these bans pretend to produce more accurate odds. They produce more trustworthy odds — odds that retail participants believe were not pre-priced by someone with a private telephone call from the locker room.

The prediction markets are now mapping that template onto event contracts. The implicit calculation: a slightly noisier price that retail trusts is worth more than a sharper price that the public believes is rigged.

There is also a regulatory dimension Hanson's argument cannot wave away. Once the CFTC determined event contracts were derivatives — a determination reinforced by CFTC Chair Michael Selig's January 2026 announcement of formal prediction-market rulemaking, and by a March 2026 staff advisory requiring real-time market-monitoring — federal insider-trading law and SEC Rule 10b-5 logic became live concerns. Exchanges have an affirmative duty to ensure their contracts are "not readily susceptible to manipulation." A platform that lets a senator trade on her own committee's vote is, on its face, in violation of that duty, regardless of what the welfare economics say.

The STOCK Act of 2012 established that members of Congress are legally prohibited from trading on material nonpublic information in equities. Whether that prohibition reaches event contracts has been ambiguous. The Moreno resolution and the Merkley bill are designed to remove the ambiguity. Whatever Hanson's purist position, the political economy is settled: Congress is not going to vote to legalize its own ability to trade on its own decisions.

What This Means for the Next Decade of Event Contracts​

The April 2026 bans look small in isolation — a handful of suspended candidates, sub-thousand-dollar fines, a Senate resolution that may or may not pass. They are not small. They are the first real microstructure choice that prediction markets have had to make at scale, and the choice will shape the category for the next ten years.

Three things follow from where the platforms have landed:

Prediction markets are pivoting from price-discovery infrastructure to consumer-grade event betting. The Hanson-era framing — that Kalshi and Polymarket are public-good price oracles for everything from elections to FOMC decisions — is being quietly retired in favor of a sportsbook framing where retail trust is the moat. The product is increasingly going to look like FanDuel-with-extra-categories, not like a futarchy laboratory.

Futarchy as a governance proposal just got harder to defend. Hanson's decision-market thesis — that we should let prediction markets pick policies because they aggregate information better than voters or experts — depends on letting insiders trade. If U.S. regulators and platforms have decided that informed trading is corruption, then any real-world futarchy implementation in this jurisdiction is structurally blocked. The thesis may move to chains and jurisdictions where the bans do not reach, but the U.S. mainstreaming case is now much weaker.

The category is splitting into two products with different physics. CFTC-regulated, U.S.-domiciled, KYC-heavy event contracts (Kalshi's lane) are accepting the integrity tax in exchange for institutional liquidity and political legitimacy. Decentralized, on-chain, permissionless markets (Polymarket-style, plus the next generation of crypto-native venues) will keep flirting with the Hanson position because they cannot enforce identity-based bans on counterparties they cannot identify. The same nominal market — "Will candidate X win?" — will trade at meaningfully different prices on the two sides.

That divergence is exactly the kind of structural inefficiency that on-chain infrastructure is built to exploit. As prediction markets fragment into regulated venues with conservative trader filters and permissionless venues with maximalist information aggregation, the arbitrage, oracle, and cross-venue settlement layers underneath them become more valuable, not less.

The Tradeoff That Was Always There​

The most honest reading of April 2026 is that prediction markets just discovered, the hard way, what every other financial market discovered in the twentieth century: pure price-discovery purity is not a stable equilibrium when retail participants and lawmakers are watching. Equity markets accepted Rule 10b-5. Sports books accepted athlete bans. Futures markets accepted position limits. Each accepted a measurable hit to informational efficiency to buy something else — public trust, regulatory tolerance, room to grow.

Hanson is not wrong that the bans cost accuracy. The 3% study suggests he may be more right, in fact, than even he realized. He is just losing the political argument, because the people deciding what prediction markets become are no longer economists. They are senators, CFTC commissioners, and platform CEOs who would rather ship a slightly worse oracle than defend a slightly better one in front of a hostile Senate committee.

The futarchy dream is not dead. It just got pushed onto chains and into jurisdictions where the question of who is allowed to know things has not been answered yet. Whether that smaller, weirder, more permissionless prediction-market sector becomes the laboratory Hanson originally envisioned — or just another offshore arbitrage venue — is the open question of the next cycle.

What is no longer open is the question that this April closed: in U.S.-regulated event contracts, the insider is not the source. The insider is the suspect.

BlockEden.xyz provides enterprise-grade RPC and indexing infrastructure across Polygon, Ethereum, Sui, Aptos, and other chains where the next generation of on-chain prediction markets are being built. Explore our API marketplace to build event-contract products on rails designed for production scale.

Sources​

Share on Twitter