In any other context, these people would be investigated, prosecuted, and punished. In the booming world of prediction markets, it’s just business as usual.
I’ve been watching this industry closely for years, and I’ve come to believe that it’s more corrosive for public life than even social media. In theory, prediction markets have genuine benefits – for information aggregation, price discovery, hedging – but these are bounded by a narrow set of conditions that political event contracts systematically fail to satisfy, and they are outweighed by the far higher costs they impose on society.
The Iran and Maduro insider-trading cases aren’t isolated incidents. The most comprehensive empirical study yet of so-called “informed” trading on Polymarket, covering more than 93,000 markets and nearly 50,000 unique wallets, found suspicious trading pairs achieving a 69.9% win rate – more than 60 standard deviations above random chance – with $143 million in anomalous profits over two years. And this almost certainly understates the problem. Skilled forecasters outperform the market, but they don’t outperform it by that much, on accounts opened days before a single event, with their largest positions placed hours before covert operations go public.
Platforms’ defenders argue that insider trading in these markets is actually desirable – that insiders have the most accurate information, and the point of these markets is to surface the truth. Banning insiders from trading, in this view, is like banning your best sources from contributing to a newspaper or the most sophisticated investors from participating in traditional financial markets. But unlike journalism and traditional finance, a prediction market is zero-sum: Every dollar won by an insider comes directly from the pocket of a schmuck who’s under the impression they’re participating in a forecasting exercise. If a few insiders dominate a market, the suckers taking the other side of their trades will keep losing and eventually run out of money or wise up and stop participating. The market will become thinner, more volatile, and accordingly less informative. Price discovery, meet adverse selection.
Sure enough, blockchain data shows roughly 70% of Polymarket traders lose money while fewer than 0.04% accounts have captured more than 70% of all realized gains, collectively pocketing an estimated $3.7 billion. Meaning the vast majority of retail bettors are chum – liquidity for the privileged and unscrupulous few who already know how the story ends.
And insider trading, serious as it is, is actually the more tractable problem. The deeper issue is what happens when insiders can bet on an outcome they have some ability to influence. A military commander choosing to retreat a day early to quadruple a month’s salary. A diplomat betting on the success or failure of the ceasefire talks he’s leading. An intelligence analyst front-running his own assessment. An elected official making public statements or public policy based on what outcome yields them highest payoff. A journalist reporting not what happened, but whatever makes him richer … or safer. These aren’t edge cases or hypotheticals. They’re the structural consequence of financializing decisions that are supposed to be insulated from market incentives. When people can profit from outcomes they can shape, prediction markets don’t surface truth as much as they create systematic incentives to corrupt it … and the people and processes that determine what counts as such.
The national security dimension is where this crosses from corrupt to dangerous. When odds on an imminent strike or an election outcome move sharply and media outlets broadcast that movement as the informed market consensus, that reported signal starts influencing how journalists frame events, what the public sees as likely and legitimate, and even how adversaries perceive intent. Iranian intelligence was almost certainly monitoring Polymarket before the February strikes. A state actor wanting to manipulate crisis dynamics could move a thinly traded geopolitical market for a few million dollars – plausibly deniable and far cheaper than mobilizing military assets – and manufacture the appearance of insider knowledge about imminent action. Cable news now quotes odds as if they were poll results. The odds become the story, never mind that it doesn’t take much to make them flip.
Kalshi says it wants to replace “debate, subjectivity, and talk with markets, accuracy, and truth.” According to its co-founder and CEO, Tarek Mansour, the company’s vision is to “create a tradable asset out of any difference in opinion.” He gets an A for honesty, but auctioning off public deliberation and reality to the highest bidder is not really the pitch he thinks it is. The differences of opinion we’ve spent several thousand years building institutions to resolve – about war and peace, about who governs and how, about whether leaders serve their country or themselves – are not differences of opinion that benefit from being turned into betting lines. Some things should remain outside the logic of the market.
Democratic governance requires a baseline of institutional faith, the belief that the people making decisions are answerable to something larger than their own interests. Prediction markets don’t just erode that faith. They financialize its erosion. Every market that opens on a political outcome is a bet that the institution overseeing it can be gamed. The more people make that bet, the more they’re right.
Most pernicious is the damage these platforms inflict on their users. We already know how addictive and financially ruinous gambling is – that’s why casinos used to be illegal most everywhere. Studies show the legalization of mobile sports betting since 2018 increased personal bankruptcy rates by around 10% in states where it took effect, tripled calls to gambling helplines, and left roughly 1 in 5 young men under 25 on the spectrum of a gambling disorder. Prediction markets follow the same playbook as these online casinos – Kalshi derives roughly 90% of its trading volume from sports bets, despite its high-brow forecasting branding. Then it extends that model into domains with higher stakes, worse information asymmetries, looser regulation, and far larger negative externalities.
When you have money riding on a nuclear detonation by the end of the decade or whether a famine reaches a certain body count, you’re no longer evaluating those outcomes and the institutions that produce them as a citizen – you’re a bettor with a directional position, rooting for whatever makes you richer. The sports betting literature is unambiguous: when there’s money on the line, even diehard fans care more about the spread than the game.
Now imagine that happening to voters. To the people who are supposed to form opinions about governance and hold institutions accountable. As Sen. Chris Murphy (D-Conn.) put it: “What happens to us, spiritually, when every moral question in this country becomes a market?” Nothing good.
I could tell you the people running these platforms are uniquely evil. But the truth is their business model – and the booming demand for their products – is the logical endpoint of America’s rotten political economy, where most young people believe the system is rigged against them and upward mobility through work and savings feels foreclosed. In this context, a bet that might 10x overnight stops looking like a vice and starts looking like the only ticket out of the permanent underclass. The platforms understand this and market to it deliberately, promising agency while delivering extraction. Kalshi’s pitch is “make your grandkids proud.” Coinbase’s prediction markets’ pitch is “take back control.” In reality, of course, prediction markets – like meme stocks and crypto and NFTs before them – don’t democratize access to wealth; they redistribute it from suckers to insiders, deepening the desperation and mistrust that drives people into the casino in the first place.
This is all happening in a country whose governance was already substantially captured. Members of Congress are allowed to hold and trade individual stocks while voting on legislation affecting those companies. The president runs a business empire and a meme coin while making executive decisions. Lobbying and campaign finance already let concentrated capital buy political and policy outcomes.
Now layer on prediction markets: anonymous accounts (Polymarket is blockchain-based, with wallets publicly traceable but nearly impossible to identify), no meaningful insider trading enforcement, and an administration whose most senior figures – Donald Trump Jr. is an adviser to both Polymarket and Kalshi – have a direct financial interest in keeping the platforms unregulated. We have created a mechanism for the politically connected to convert their access into private profit. At least lobbying buys outcomes on behalf of institutions with broader interests and accountability. Prediction market insider trading is purely personal: a single anonymous wallet, in or out of the money, accountable to no one.
None of this is to say that prediction markets have no legitimate use case. Hayek was right that markets aggregate dispersed private knowledge better than any centralized mechanism. On apolitical, technical questions – where will inflation land, will the Fed cut rates – prediction markets have been found to perform about as well as the best professional forecasters, sometimes better. Beyond information aggregation, there’s also a genuine hedging argument: multinationals exposed to regime change, energy majors exposed to shipping-lane disruptions, a manufacturer exposed to import tariffs – these are real risks that well-designed event contracts could in principle help insure against. But both cases require conditions that fail systematically on political questions – conditions that existing platforms were deliberately built to violate.
The wisdom of crowds requires independent participants, dispersed information, and no ability to move the outcome. Ask crowds to predict outcomes that are politically charged, emotionally loaded, or controlled by a small number of people with nonpublic information, and the conditions for wisdom collapse. Opinions stop being independent and start reflecting tribal alignment or the positions of a few large “whales” whose bets materially shift the odds.
The hedging logic, meanwhile, only works when the hedger can’t influence the outcome, information asymmetries are manageable, and participants are identifiable. Blockchain infrastructure, pseudonymous wallets, and offshore jurisdiction make those conditions impossible to meet. The same anonymity that might serve a legitimate hedger serves the defense contractor who knew about the Iran strike 71 minutes early. You cannot build a legitimate risk market on infrastructure designed to be unaccountable.
Only after mounting scandals did both major platforms announce enhanced rules – banning trading on stolen confidential information, prohibiting trading by anyone who can influence an outcome, and blocking political candidates from wagering on their own campaigns. But when Kalshi’s CEO says the long-term vision is to “financialize everything” and Polymarket’s CEO describes insider trading as “sort of an inevitability” with “lots of benefits,” the limits of self-policing are obvious. Sen. Todd Young of Indiana, the lead Republican sponsor of bipartisan legislation to restrict such bets, identified the deeper problem: prediction markets create “incentives or temptations that those who hold positions of public trust might harbor,” and no platform terms of service can legislate that away.
Which is why the growing governance momentum matters, however partial. California Governor Gavin Newsom signed an executive order last week barring state officials from using nonpublic information to profit on prediction markets. Multiple bills have been introduced in both chambers recently – barring the president, members of Congress, and senior executive branch officials from trading on prediction markets, prohibiting event contracts tied to war and assassination, expanding CFTC and state enforcement authority. Red and blue states alike are moving within their own jurisdictions. Whether any of it survives congressional gridlock and court challenges is uncertain.
Institutions that trade in trust can’t afford to wait to find out. That’s why two weeks ago, Eurasia Group banned all staff and their families from betting on prediction markets. No incident prompted the decision. Our people are a principled lot – hence why they work here rather than at the firms that get paid to produce conclusions. But the structural incentive is there regardless of individual ethics. As analysts, we often have access to nonpublic information about tradable events. We hold confidential client information about how major institutions are thinking about and positioning around those same events. And we sometimes have some degree of influence over how outcomes unfold. Trading on our analysis wouldn’t make us better at our jobs – it would just give us a hidden stake in the outcome and shroud our work in suspicion. We’re already plenty accountable: our calls are falsifiable, time-bound, and transparent to our clients, and we face real reputational and commercial consequences when we’re systematically wrong. That’s a healthier incentive than an anonymous financial position on a binary outcome, because it rewards understanding the world accurately over time – and changing our minds when the facts change – rather than being right about this contract, by this date, according to these resolution criteria.
The ban can’t prevent violations – that’s what our hiring process is for. It’s also not a substitute for regulation – or for society deciding that some things simply shouldn’t be tradable. An internal policy governs what happens inside our walls; it doesn’t govern the defense contractor, the financial analyst, or the elected official on the other side of the market. What it does is make clear where we stand and ensure that our clients, sources, and readers never have reason to wonder whether our analysis is chasing the truth as we see it or chasing a payout.
Will others follow? I wouldn’t bet on it.
