Hi there, argmin readers. Today’s post is a live blog of Class 19 of my graduate seminar “Forecasting: A Critical Retrospective.” In this class, we try to make sense of prediction markets. Good luck to us! The syllabus and list of past posts are here.
On May 16, 2008, in the Policy Forum of Science magazine, two dozen notable scholars called1 on the Commodity Futures Trading Commission (CFTC) to create a safe harbor for prediction markets. In their view, “vibrant, liquid prediction markets” would improve decision-making by “helping businesses make better investment decisions,” “helping governments make better fiscal and monetary policy decisions,” and helping “private firms and public institutions in managing economic risks.” The predictive magic came from the unreasonable efficiency of markets:
“Because information is often widely dispersed among economic actors, it is highly desirable to find a mechanism to collect and aggregate that information. Free markets usually manage this process well because almost anyone can participate, and the potential for profit (and loss) creates strong incentives to search for better information.”
Just two months earlier, on March 17, 2008, the Federal Reserve had convened an emergency meeting, its first in thirty years, to guarantee the failing loans of Bear Stearns and prevent it from collapsing into bankruptcy. The world would quickly plummet into the worst economic recession since the Great Depression. It was the efficiency of markets in action.
Fast forward to 2026, and prediction markets have parasitically clawed their way into ubiquity. Trading volume on sketchy legalized gambling sites Polymarket and Kalshi grew fivefold this year. These companies are also aggressively lobbying the CFTC, but without the technocratic idealism of the 2008 policy wonks. Kalshi and Polymarket are gambling websites, and everyone knows it. But they are using old academic arguments to skirt state gambling laws. The CFTC regulates futures markets. The states regulate gambling. If the CFTC says prediction markets are futures markets, then they are protected from the authority of the states. The Trump family is heavily invested in crypto and prediction markets and is pushing hard to bend the CFTC to their will to enshrine the legal status of the companies.2
The CFTC and its regulatory predecessors have always been tasked with carving out “safe harbors” for futures markets because you have to tie yourself into neoliberal knots to explain how derivatives markets are not gambling. Like most financial instruments, futures markets rest on a simple idea that feels reasonable, but it spirals out of control once you let the invisible hand of the market get to work.
A company that relies on corn production might want to set a price for corn in advance so they can manage risks from volatile harvests. They make a contract with some farm producers before the season, agreeing upon how much they’ll pay per bushel at the end of the season. If there’s a rough harvest, the market price will be considerably higher than the contract price, and the buyer gets their corn at a discount. On the other hand, if the corn harvest is plentiful, the contract price will be higher than the market rate. However, this isn’t the end of the world, because they can buy corn at a cheaper rate from another farmer as supply is abundant. Properly balancing corn on futures contracts and corn at actual prices leads to an expected cost the buyer feels comfortable with. The futures contract serves as a risk management strategy, allowing producers and buyers to hedge against an uncertain market.
Now, a market fundamentalist would argue that for everyone to get the best prices for their businesses, a bigger, more liquid market should be established to determine the right price. People should be able to sell their contracts to each other to dial in the right price as new information emerges. But this ideal world of making all sorts of contracts with different producers and buyers is physically messy. If you wanted to execute all of these deals, you’d have to ship the corn all over the place at harvest time, creating a logistical nightmare. So instead, people created a market for these futures, where you’d sell your futures contract before it ever gets executed. No corn moves anywhere, but money still changes hands. The risk management stuff still works out for all involved, but no one is buying any actual corn in the corn market.
And shoot, if you’re not buying actual corn, why not buy virtual everything else too in these markets? You’re buying a contract on the price of corn, not the corn. You could imagine buying a contract on the price of a financial instrument, like a stock price or an interest rate. This would help you manage risk in an investment portfolio. After its establishment in the 1970s, the CFTC began liberally including all sorts of futures in its “no, this is actually not gambling” bucket. Sure, you’re not buying anything in these markets except contracts on future prices. But it was of economic value to allow such speculation. The CFTC still discouraged derivative contracts on “events” that didn’t have a clear economic purpose.
Meanwhile, egghead economists argued the CFTC should relax. Any event can be securitized, and this securitization is good for everyone. Ken Arrow and Gerard Debreu won the Sveriges Riksbank Prize in Economic Sciences in Memory of Alfred Nobel for this sort of thinking. Inspired, economists in Iowa started running fake markets to predict the outcome of the Iowa caucuses. And people like Phil Tetlock started to run competitions with IARPA to improve forecasting geopolitical turmoil. But prediction markets add another layer of derivatives. Betting on events lets decision makers discover probabilities they can plug into expected-value calculations, so they know how to price commodity futures contracts. I guess?
I am personally fascinated by this pattern of derivatization in finance. Derivatives feel sensible for risk management, but require creating a bizarre abstraction of the object you’d like to hedge against. These derivatives aren’t necessarily gambling, but derivatives on derivatives are always gambling. There’s no way to avoid the cascade into speculation once you financialize reality.
Prediction markets are a hilarious caricature of neoliberal economics. In prediction markets, everything is gambling, and everyone knows it. No matter what the rationalist disciples of Tetlock tell you, there’s no sensible defense of the prediction markets. They don’t produce probabilities in any reasonable sense, in theory or in practice. They don’t inform decision-making. They are not “efficient,” as no markets are efficient. The prediction market companies exist entirely for gambling. And gamblers love them.
If you want to defend them, you need to embrace a certain style of hyper-rational thinking that we’ve mistakenly normalized. We are told that the predictions are good because markets are the ultimate information aggregators, and the information is good because people in markets have skin in the game. But who takes advice from degenerate gamblers?
I mean, look at this author list. The argmin nightmare blunt rotation. Kenneth Arrow, Robert Forsythe, Michael Gorham, Robert Hahn, Robin Hanson, John Ledyard, Saul Levmore, Robert Litan, Paul Milgrom, Forrest D. Nelson, George R. Neumann, Marco Ottaviani, Thomas C. Schelling, Robert J. Shiller, Vernon L. Smith, Erik Snowberg, Cass Sunstein, Paul Tetlock, Philip Tetlock, Hal R. Varian, Justin Wolfers, Eric Zitzewitz.
For more on these developments, read John Cassidy’s reporting “This Is What Trumpian Self-Dealing Looks Like” in The New Yorker.
I might ask the question the other way: what part of gambling isn't prediction markets?
It seems clear that gambling has a definition here but it wasn't until we basically let a sports book into every phone that this took off.
Thank you for using the correct name of the Sveriges Riksbank Prize in Economic Sciences in Memory of Alfred Nobel.