The Data Behind the Theory:
What the WSJ’s Prediction Markets Analysis Means for the CFTC’s Derivatives Classification
De Silva Law Offices, LLC
The Wall Street Journal just published the most comprehensive empirical analysis to date of who wins and who loses on prediction markets. The findings should be considered by every regulator, legislator, and jurist currently deciding the future of these platforms.
Two weeks ago, a piece published on this site argued that prediction market prices are not the product of crowd-sourced truth discovery but the byproduct of a wealth transfer from unsophisticated retail participants to professional algorithmic trading firms. That argument drew on Sanford Grossman and Joseph Stiglitz’s impossibility theorem, on Albert Kyle’s 1985 microstructure model, and on the structural logic of how electronic order books actually work. It was theoretical. But on May 3, 2026, the Journal supplied the real world data validating the theory.
The Journal’s analysis covers 1.6 million Polymarket accounts and over 35,000 completed mention-market contracts on Kalshi. The concentration of profits, the systematic mispricing of contracts, and the role of institutional market makers all appear in the data exactly as the microstructure literature would predict.
The timing is significant. The CFTC’s March 2026 Advanced Notice of Proposed Rulemaking on event contracts closed for comment last week. State challenges to Kalshi’s Designated Contract Market status are pending in multiple jurisdictions. Congress has several bills in play. The empirical record available to all of these decision-makers changed materially on May 3.
Three Findings
The Journal’s analysis produced three data points that deserve separate treatment. Each maps onto a specific element of the regulatory framework.
Profit concentration.
Of the 1.6 million Polymarket accounts the Journal analyzed, 0.1 percent captured 67 percent of all profits. Fewer than 2,000 accounts netted a combined total approaching half a billion dollars. More than 70 percent of users lost money. On Kalshi, the ratio is roughly 2.9 unprofitable users for every profitable one, according to the company’s own spokeswoman.
None of this is surprising if you understand Kyle’s model. Informed traders extract profits from noise traders. Market makers take a cut for intermediating. The profits do not distribute evenly. They concentrate in the hands of whoever has superior information, superior technology, or both. A market in which 0.1 percent of participants capture two-thirds of all gains is not aggregating the wisdom of a crowd. It is sorting participants by informational advantage and transferring wealth accordingly.
The Journal’s profitability chart tells the rest of the story. It plots the share of profitable traders by trading frequency percentile. At the top 0.1 percent by frequency, roughly half are profitable. Below the sixtieth percentile, fewer than half are. The relationship is monotonic. Trade more, and the odds of winning improve. Trade less, and they deteriorate. That pattern is the signature of a market dominated by professional participants who trade constantly and profit incrementally, at the expense of casual participants who trade intermittently and lose in chunks.
Mention market mispricing.
The Journal analyzed over 35,000 completed mention markets on Kalshi. These are contracts that pay out based on whether a public figure says a particular word during a public appearance. The Journal found that “yes” trades priced at a 50 percent implied probability paid out only about 40 percent of the time.
Lets consider that again. A contract the market prices as a coin flip is actually worse than a coin flip, systematically and persistently. The retail participant buying at the listed price is overpaying relative to the contract’s actuarial value. The Journal calculated that a retail trader buying mention-market contracts at the first available price faces an expected loss of roughly 11 percent of the amount wagered. Research from the University of Nevada, Las Vegas indicates that this expected loss rate is worse than most slot machines on the Vegas Strip.
Professional traders told the Journal they avoid mention markets entirely. The outcomes are too unpredictable, and even expensive data cannot produce a reliable edge. So who is on the other side? Retail participants, drawn in by social media influencers who livestream their trades during events. Bank of America analysts noted in an April report that mention-market livestreams were designed to boost engagement and brand awareness for Kalshi.
The regulatory implication is specific. A market that is systematically miscalibrated is not performing price discovery. It is performing price distortion. The contracts are not converging on the true probability of the event. They are converging on a price that is too high, because the buy side is dominated by retail participants exhibiting long-shot bias and the sell side is populated by professionals who understand the actuarial value. That is an extraction mechanism, not an information-aggregation mechanism.
Institutional counterparties.
Susquehanna International Group signed on as Kalshi’s first major institutional market maker in 2024 and is believed to trade hundreds of millions of dollars through the platform each week. Jump Trading is active on both Kalshi and Polymarket. Citadel Securities has said publicly that it is watching the space. A firm founded by college students told the Journal it spends over $200,000 annually on live data feeds, AI coding agents, and servers, executing tens of thousands of algorithmic trades per day. It has turned $1,000 into seven-figure profits.
One of the professional traders profiled, a former poker player and trained statistician, places 60 trades per minute on Kalshi and updates his bids and asks 30 times per second. On the other side of those trades are people like John Pederson. Pederson is 33, a former line cook who took out a variable-interest loan to start betting on prediction markets. He ran $2,000 up to $41,000, bet it all on a single mention-market contract, and lost everything. He is now living in a homeless shelter in Detroit.
The Susquehanna connection deserves particular attention. Jeff Yass, the firm’s co-founder, said on a 2020 podcast that the key to sports betting, poker, and options trading is making sure you are always betting against someone less capable. He described his role supporting what would become prediction markets as a mission. Susquehanna’s trading profile on Kalshi is private. The firm declined to comment. How much money it is making is unclear. What is clear is that a multi-billion-dollar quantitative trading firm with decades of experience in options market-making, and a co-founder who openly describes retail counterparties as the business model, is now the dominant liquidity provider on a CFTC-regulated prediction market exchange.
What This Means for the Classification
The CFTC’s treatment of event contracts as derivatives rather than gaming products rests on a specific intellectual claim: that prediction markets perform price discovery by aggregating dispersed information into prices reflecting the true probability of future events, and that this function distinguishes them from gambling. The April 20 piece argued that this claim is load-bearing at the precise point in the regulatory framework where the derivatives classification is made, under CEA Section 5c(c)(5)(C), which prohibits event contracts that involve “gaming.”
The Journal’s data makes that claim harder to sustain. Consider what the CFTC would need to argue to maintain the current classification in light of the new evidence.
It would need to argue that a market where 70 percent of participants lose money is performing a valuable public forecasting function. It would need to argue that systematic mispricing in mention markets, where contracts routinely trade above their actuarial value, constitutes price discovery rather than price distortion. It would need to argue that the concentration of 67 percent of all profits in 0.1 percent of accounts reflects the healthy functioning of an information-aggregation mechanism rather than the healthy functioning of an adverse-selection mechanism. And it would need to argue that a regulatory framework designed for institutional derivatives counterparties is appropriate for a user base that includes people taking out loans to bet on whether a rapper will say a particular word on late-night television.
Each of those arguments was already difficult on theoretical grounds. With data, they become untenable.
DCM Obligations
A Designated Contract Market has self-regulatory obligations under the Commodity Exchange Act. Two are directly implicated here.
Core Principle 4 requires a DCM to prevent manipulation, price distortion, and disruptions of the delivery or cash-settlement process through market surveillance, compliance, and enforcement. The systematic mispricing in mention markets raises a question under this principle. If a category of contracts consistently trades above its actuarial value because retail participants exhibit long-shot bias and professional traders refuse to participate, is the exchange meeting its obligation to prevent price distortion? Nobody is spoofing the order book or painting the tape. But the distortion is real, persistent, and measurable. The Journal measured it.
Core Principle 12 requires a DCM to protect market participants. Pederson’s story is a case study in the absence of that protection. A financially vulnerable individual, recovering from a car accident and running out of money, took out a loan and placed it on a prediction market. The platform’s marketing materials implied that ordinary people could profit. A TikTok ad featured a woman claiming she had earned two years of rent. Pederson parlayed small wins into a large stake and then lost it all on a single mention-market bet whose settlement rules were not immediately visible on the platform’s interface. He did not see them. Kalshi has since updated its interface to make market rules more apparent.
State gambling law would have provided Pederson with loss limits, self-exclusion mechanisms, mandatory disclosures about the odds of winning, and restrictions on the kinds of advertising directed at him. The federal derivatives framework provides none of these protections. Whether that trade-off is justified by the price-discovery benefits the CFTC has cited is the central question. The Journal’s data suggests it is not.
The ANPRM and the State Cases
The CFTC’s March 2026 ANPRM asked whether the Commission should amend or issue new regulations governing prediction market event contracts. Among other things, it solicited comment on which categories of event contracts should be prohibited as against the public interest.
Mention markets are a strong candidate. Professional traders refuse to participate because outcomes are too unpredictable. Retail participants systematically overpay relative to actuarial value. Expected losses exceed those of most casino slot machines. The platform has used influencer marketing and livestreaming to drive engagement. A contract category with these characteristics is difficult to defend as serving the public interest under any reasonable construction of that standard.
The state cases benefit from the Journal’s data in a different way. New Jersey, Nevada, Ohio, and other states have argued that Kalshi’s event contracts are functionally indistinguishable from gambling products and should be regulated under state law rather than preempted by federal derivatives regulation. Their core argument is that the “price discovery” rationale separating a derivative from a wager does not hold up when you look at what is actually happening on these platforms.
The Journal just looked. What it found is a market where most people lose, a small number of professionals win, the contracts are systematically mispriced in the professionals’ favor, and the platforms are marketing aggressively to the people who lose. That description fits a gambling product more comfortably than it fits a derivatives market. The states will use it.
The Platforms’ Responses
Kalshi’s spokeswoman told the Journal that many financial markets exhibit similar patterns of wealth concentration, and that more users make money on Kalshi than through day trading or traditional sportsbooks. She said Kalshi no longer runs the “pay my rent” advertisement. She acknowledged that mention markets face long-shot bias but argued they are not representative of the platform’s overall pricing. She added that Kalshi’s own analysis showed mention markets were priced more accurately in the four hours before an event.
That four-hour qualification is revealing. If a contract is only accurately priced in its final hours, it was mispriced for the much longer period during which most retail trading occurs. A market that converges on accuracy only at the last moment is not performing continuous price discovery. It is performing a slow correction of accumulated retail overpayment.
Polymarket declined to comment. The Journal disclosed that Polymarket has a data partnership with Dow Jones, the Journal’s publisher, and stated that it used only publicly available data for its analysis.
Conclusion
The April 20 piece on this site asked a question: Do prediction markets discover truth, or do they redistribute information? The Wall Street Journal answered it with 1.6 million data points.
The answer is redistribution. It is overwhelming, systematic, and concentrated. The retail participants whom the platforms market to most aggressively bear the cost. The professional firms providing liquidity collect the proceeds. Prices that emerge from this process carry some information, as prices in any traded market do. But that information content does not justify a regulatory classification that strips away consumer protections designed for exactly the population that is losing.
The CFTC has an open rulemaking. The states have pending litigation. Congress has multiple bills under consideration. All of them will benefit from the more accurate empirical picture the Journal has now provided. The theoretical argument was already difficult for the current framework. The data makes it harder.
De Silva Law Offices, LLC is a Chicago-based boutique firm specializing in CFTC and NFA regulatory matters, securities law, derivatives, and event contract compliance. The firm advises platform operators, introducing brokers, commodity pool operators, commodity trading advisors, and individual market participants on the regulatory architecture of prediction markets and related derivatives, including designation and registration, contract review, enforcement defense, and regulatory classification analysis.
Sources
“Why Almost Everyone Loses Except a Few Sharks on Prediction Markets,” Wall Street Journal (May 3, 2026).
R Tamara de Silva, “Do Prediction Markets Discover Truth, or Redistribute Information?,” De Silva Law Offices (April 20, 2026).
CFTC, Advanced Notice of Proposed Rulemaking: Event Contracts (March 2026).
Sanford Grossman and Joseph Stiglitz, “On the Impossibility of Informationally Efficient Markets,” American Economic Review 70 (1980).
Albert Kyle, “Continuous Auctions and Insider Trading,” Econometrica 53 (1985).
Nizan Geslevich Packin and Sharon Rabinovitz, “Prediction markets as a public health threat,” Science (April 2026).