A few years ago, I met the economist Robin Hanson, one of the architects of prediction markets. Over lunch, Hanson made a remark that I have not been able to shake: stock markets punish insider trading; prediction markets reward it.
The line sounded, at first, like an admission of something nefarious. Insider trading is normally frowned upon, and even punished, because it can distort markets and disincentivize investing by outsiders. But after thinking about it for a bit, I realized that insider knowledge is the whole point of prediction markets. Unlike with the conventional stock market, the goal of prediction markets is to incentive people to broadcast their insider knowledge. Their winning bets are a way of making that information more widely available.
That memory was made new again for me by recent news that the Department of Justice is charging an Army soldier with five felonies for allegedly betting roughly $33,000 on Polymarket that a raid to capture Nicolás Maduro would happen. The raid occurred and the soldier cashed out around $400,000. Kalshi, another prediction market, suspended the accounts of candidates for office who bet on their own races. Both platforms are scrambling to bar insiders of various kinds from trading on contracts tied to their work.
Apparently, almost no one in the policy world has absorbed Hanson’s insight: that prediction markets are useful because they make knowledge more widely available. That worthy goal is impeded and disincentivized when people who provide such knowledge are banned from the platform and punished by the government.
To see why insider trading on prediction markets is a feature rather than a bug, it helps to begin with a question about the meaning of “knowledge”. Most of what any of us takes ourselves to “know” about the world is acquired through testimony from other people who, presumably, know better than we do. Whether the topic is monetary policy, the safety of a vaccine, the timing of a military operation, or the likely outcome of a Senate race, nearly every bit of knowledge we have comes from others who are better placed to know than us.
The social character of knowledge means that the structure of incentives facing those who hold information matters as much as the information itself. The used car salesperson typically knows more about the car than the buyer does. His expertise is real, but his incentive to faithfully convey that information is weak. The public expert and the salesperson differ less in what they know than in why they are talking. Public experts get rewarded when they share their expertise, like a geologist who wins acclaim, a more prominent appointment, or perhaps a raise after they successfully predict a volcanic eruption. The salesperson lacks such incentives, at least if sharing too much of what they know would impede the sale, as when they know that the car they’re selling is a lemon..
Almost every problem of testimony, from cable news to peer review to political speech, reduces to some version of this incentive asymmetry.
Most of our institutions for producing public knowledge handle the incentives problem badly. Politicians and bureaucrats can face career penalties for being right at the wrong time. Perhaps the most famous example is Winston Churchill, who warned for over a decade about the threat that Hitler posed to the world. They rarely face any penalty for being wrong when they agree with the herd. Consder the many politicians who favored the War in Iraq with little consequence to their careers. To give another example, scientists are supposed to be immune to public pressures. Yet they work inside reputational economies that reward consensus more than calibration.
Journalism, at its best, helps close the gap. But the typical reporter has limited ability to verify the claims of official sources or to compel disclosure from those who prefer to keep quiet. The result is an epistemic environment where much of what is true is known only to a small number of people who have no professional reason to share it and often have strong reasons keep quiet. Even when we think we’re well-informed, we’re really in the dark.
Prediction markets, at their best, are an institutional fix to that problem. In effect, they pay people to come forward with what they know when almost no one else knows yet. If someone has information that leads them to believe the price for a certain outcome in the prediction market is wrong, they can profit by trading until the price is right. For example, throughout the 2012 presidential campaign, Intrade had Obama’s reelection probability fluctuating, sometimes dipping to around 55-60% during Romney’s post-debate bounce. Nate Silver’s model at FiveThirtyEight, which was essentially private analysis unavailable to most traders, consistently showed Obama at 70-80%+. Sophisticated traders who trusted Silver’s methodology over the conventional pundit wisdom — which was treating the race as a tossup — bought Obama contracts at the deflated prices. When Obama won decisively, they profited.
This is how prediction markets convert private knowledge into a public good. They function less like casinos than like a kind of epistemic commons—an open-source library for cutting-edge knowledge that would otherwise be locked away. To give a historical example, on the day of the Challenger explosion in 1986, the stock market identified the culpable company, Morton Thiokol, maker of the faulty O-rings, within minutes of the disaster. Economists Maloney and Mulherin documented how Thiokol’s stock fell sharply and disproportionately while the other three implicated contractors recovered within the same trading session. Someone with private technical knowledge—engineers aware of the O-ring cold-weather concerns, or perhaps suppliers or contractors—sold Thiokol stock and price movement aggregated that dispersed private information almost instantly. The Presidential Commission took months to reach the same conclusion the market reached in minutes.
That is why the panic over “insider trading” on these platforms is misplaced. Consider trades that have most alarmed regulators: a soldier with classified knowledge of an upcoming raid bets that the raid will happen, or an anonymous Polymarket trader identifies four pardons before they are announced, walking away with around $300,000. In these cases, the trade is profitable because the trader is moving the market price toward the truth. Thanks to these insiders, prediction markets are demonstrably faster and (often) better than the press at bringing private information to light.
Compare this to ordinary financial markets, where similar activity is treated as a serious crime. Hanson rightly notes that insider trading is, in practice, rampant in equities markets. When a public company makes a major announcement, a substantial fraction of the price move happens before the news breaks. The SEC successfully prosecutes only a small share of those trades.
The rule itself, moreover, is a relatively recent invention. The original twentieth-century prohibition was narrow—corporate officers trading on their own firms’ information—and not unreasonable as a corporate-governance restriction. About fifteen years ago, the Commodity Futures Trading Commission generalized the prohibition to cover, roughly, anyone who had ever promised to keep a secret. That expansion changed insider-trading law from a narrow rule about fiduciary duty into a sweeping obligation to protect the secrecy of others.
None of this is to deny that organizations have legitimate interests in keeping certain secrets. A soldier betting hundreds of thousands of dollars on the timing of a classified mission is not a hard case. That is a clear operational risk and rightly punished. Yet the appropriate remedy is internal to the military, not a wholesale ban on insider participation in prediction markets.
And, of course, markets can be too thin to be accurate, too easy to manipulate, or vulnerable to obvious conflicts of interest. For example, on October 23, 2012, two weeks before the presidential election, a single anonymous trader on Intrade placed a massive series of bets on Romney winning, moving his perceived win probability from 41% to over 50% within hours. Betfair, which had far greater volume, barely moved. The market was thin enough that one actor with sufficient money could single-handedly shift the headline number that major media outlets were reporting as genuine crowd wisdom. The price eventually corrected, but the episode illustrated all three failure modes at once: the market was too thin to resist manipulation, too lightly monitored to identify or stop it, and vulnerable to an obvious conflict of interest.
These are real problems worth addressing, but they do not indict prediction markets as institutions. Markets at their best are still better than the alternatives at theirs.
What about the people on the other side of these trades, those who lost money to insiders? Hanson has a line he gives his finance students: when you sit down at a poker table, look around and find the fool; if you cannot find the fool at the table, get up and go, because it is you. The advice sounds harsh, but the point here is that a market that pays insiders for the truth is, by construction, a market that ordinary participants should approach with humility. Prediction markets are not for amateurs. You should only bet when you have insider knowledge, not when you simply have a strong opinion or instinct..
The people most worth listening to are often the ones with the most to lose by being wrong. Applied without discrimination, insider-trading rules push out exactly the people that prediction markets exist to attract—those with genuine knowledge and a real stake in the outcome. So while some narrow prohibitions make sense, like those for operational security, the wholesale extension of these rules to anyone holding non-public knowledge defeats the purpose. The best prediction markets will be those that sanction the trades that cause real harm and protect the ones (based on benign insider knowledge) that move prices toward the truth.
About The Author
