Should You Invest in Prediction Markets?

You’re reading Dispatch Markets, a weekly dive into the forces driving economic growth—and those holding it back—featuring Scott Lincicome, Kyla Scanlon, Karl Smith, Marian Tupy, and Adam Ozimek. Prediction markets are inescapable. If you open up your brokerage app, you will likely see them: Robinhood has a prediction markets hub, and so do Coinbase, Webull, Interactive Brokers, and Tastytrade. Many of those trades done on the apps flow to Kalshi, a federally licensed exchange that now pays interest on idle cash that users have in the app, like a brokerage would. The message is pretty clear: Prediction markets are a new kind of investment, and you should jump in if you want to make a lot of money. So, should prediction markets be a part of your investment portfolio? Well … no. Bad Bets Jackson Merrill of the San Diego Padres runs to center field in front of a Kalshi ad before a game against the New York Yankees at Petco Park on Friday, Sept. 4, 2026 in San Diego. (Photo by Meg McLaughlin/San Diego Union-Tribune/ Getty Images) When you buy a stock, you own a small piece of a company. Think of it like an apple—every slice of the apple is a share that you buy to own. The whole apple is the company. The company makes things, sells them to people, earns money in doing so, and eventually, some of that money comes back to you in the form of dividends or a higher price when you sell the stock. When you buy a bond, you are lending money to some entity, maybe the U.S. government or Amazon, and getting paid interest in return. With both shares and bonds, there are underlying fundamentals that produce value over time. But an event contract, which is what prediction markets trade, has nothing underneath it that produces value over time. It’s a binary. Say you think the Kansas City Chiefs will win Sunday and someone else thinks that they won’t. The contract pays $1 to whoever is right and $0 to whoever is wrong. There is no real productive value, no real earnings. It’s just taking money from one person and giving it to another, with the exchange taking a cut for arranging that transaction. On Kalshi, the fee is 3.5 cents for every dollar the buyer puts in on a 50-cent contract, with proportionally higher fees for long shots. Say you put $10 on the Kansas City Chiefs and the market has the game as a coin flip (other traders decide the odds here, not Kalshi). A contract will pay $1 if they win and costs you 50 cents. Your $10 buys 20 contracts. Kalshi’s fee is 7 percent of the contract price times the chance of losing on every contract—so 20 x $0.50 x 0.50 x 0.07, which comes to 35 cents. So $10.35 leaves your account: $10 for the contracts and 35 cents for Kalshi. If the Chiefs win, each contract pays $1, and you collect $20. That’s a profit of $9.65 after the fee. If they lose, you collect nothing and are out $10.35. Now do that 100 times. Say it is a coin flip, and you win 50 of them, collecting $20 each. You get $1,000, but you paid $1,035. You are down $35, which is 100 fees of 35 cents each. Even if you’re right exactly as often as the odds say, you lose. Advertisement Stay ahead of the policies shaping free enterprise and impacting American business. Get the U.S. Chamber’s free newsletter for insights on the economic policy, workforce trends, and regulatory landscape that affect businesses and markets. By subscribing you agree to receive communications from the U.S. Chamber of Commerce. Michael Mauboussin, the head of consilient research at Morgan Stanley Investment Management, has a test for asking whether something requires skill: Can someone lose on purpose? A chess master can lose whenever he or she wants. But a roulette player cannot lose on purpose because the wheel spins due to physics. Mauboussin found that investing sits closer to the luck end of the scale than most people think. Picking a stock is closer to a slot machine than most people think, but a stock held long enough is expected to pay you, whereas a slot machine is expected to rob you. For a person clicking buy, a prediction market behaves like a slot machine. In the example above, it’s truly a coin flip. And a coin-flip example is the best case here, as ordinary users tend to do a lot worse. A team of economists led by Karl Whelan of University College Dublin studied more than 300,000 Kalshi contracts using data through early 2025 and found that long shots are overpriced. Contracts costing under 10 cents lose more than half of the money that people put into them. They also found that people who post offers on the website lose about 10 percent on average across all contracts, whereas people clicking to accept the offer lose almost 32 percent, with the worst of it on long shots. The Wall Street Journal looked at 1.6 million accounts on Polymarket, Kalshi’s main rival, and found that 67 percent of all the profit on the platform went to 0.1 percent of accounts. That’s fewer than 2,000 traders splitting roughly $500 million, while 70 percent of users are losing money. Kalshi’s spokeswoman told the Journal that there were 2.9 losers for every winner on its platform in a recent month. The Roosevelt Institute, a progressive think tank, went through 400 million Kalshi trades and estimated that everyday users had lost more than $500 million since launch, most of it on sports. Kalshi disputes this number, arguing that the study mislabeled “ordinary users” and “professional users.” Kalshi says that more of its users make money than they would at a sportsbook or at day trading, but that doesn’t mean that most of them aren’t still losing. So who wins? The Journal profiled a trading firm run by college students that spends more than $200,000 a year on live data feeds and servers to fire off tens of thousands of Kalshi trades per day. Prediction markets are becoming increasingly institutionalized, and someone trading on their phone at halftime has no edge against that. So are prediction markets all bad? No. Again, Kalshi doesn’t set odds the way a sportsbook does. Prices come from a running list of what other traders are willing to pay, and the exchange itself collects a fee on each trade. The fee looks small, and on big events it certainly is. Securities analyst Jordan Bender found Kalshi’s all-in pricing beat the sportsbooks during March Madness and the FIFA World Cup. None of this makes prediction contracts an investment, but there is a real use case in hedging. Let’s say you’re a homebuilder and your year depends on the Federal Reserve cutting rates to lower borrowing costs and to get people back in the housing market. You could buy contracts on Kalshi’s Fed-meeting market that pay out if the Fed doesn’t cut, so if things don’t go your way, you still get money from the contract. It’s the same way a farmer would use corn futures. But. The company does run its own affiliated market maker, Kalshi Trading, which posts prices and takes the other side of user trades alongside firms like the Wall Street trading firm Susquehanna. At least three class-action lawsuits argue that this does qualify as a “house.” Kalshi says the affiliate isn’t profitable and this is “a common and regulated practice,” but it’s not what “peer to peer” looks like. Better prices also don’t mean better outcomes, as Bender found by using data from a bet-tracking app. He found that the median prediction market user lost 8 percent between July 2025 and March 2026, which is worse than the 5 percent sportsbook bettors lost. The only group that made money was traders moving more than $500,000, and they made it on prediction markets, not sportsbooks, which would have banned them. According to CoinDesk reporting, two professional bettors said on a Citizens JMP call that the appeal of prediction markets is that retail users, people trading on their phones during the game, provide the liquidity. And no one is hedging. Prediction markets got their start as places to bet on elections, but now they are mostly a sportsbook. According to the 9th Circuit Court of Appeals, which ruled against Kalshi last month in a suit filed by the state of Nevada, sports made up more than 90 percent of Kalshi’s trades and 95 percent of its revenue in 2025. And since sports betting is regulated by the states, the states have stepped in. Kalshi got the right to run election markets in 2024, when a federal judge in Washington, D.C., ruled that betting on elections did not constitute “gaming” under the law that governs commodities exchanges. Regulators appealed, then dropped the appeal in May 2025 under new leadership. Kalshi had started listing sports contracts that January, and by March a bunch of states were trying to get the platform to stop, as sports betting is usually regulated state by state. But Kalshi argued it is an event-contract company regulated by the Commodity Futures Trading Commission (CFTC), and that federal licensing overrides state gambling laws. The CFTC agrees and has sued Illinois, Connecticut, and Arizona, which have sent cease-and-desist letters to Kalshi and other prediction market operators. President Donald Trump posted in May that the CFTC should be the only regulator. The courts are split. More than a dozen states are in litigation with Kalshi, some as plaintiffs and some because Kalshi sued them first. A bipartisan group of almost 40 attorneys general has filed a brief to back the states. On April 6, the 3rd Circuit sided with Kalshi and blocked New Jersey. On August 28, the 9th Circuit sided with Nevada, ruling that a contract on who wins a football game is betting. The opinion’s first sentence quotes Kalshi’s ad that says “the first app for legal sports betting in all 50 states.” The Supreme Court will likely have to get involved, and the legal system is tangled up with numerous lawsuits and suggested bans. Gambling is a state matter, and the 9th Circuit’s logic is that sports contracts offered by the prediction markets are gambling so they should be regulated as such. A federal agency’s decision to bless a product should not strip Nevada of its right to regulate sports betting inside Nevada. States should have a say. So how do we fix this? The first fix is to cut the parlays. Kalshi calls them “combos” in which you string together several bets, so the Chiefs winning, the game score totaling more than 47 points, the quarterback throwing two touchdowns, an eagle flying across the sky. It only pays if every single leg hits. The odds tend to get worse with the more legs that you add, and the Roosevelt Institute found that those accounted for more than $100 million of everyday user losses (Kalshi’s accuracy study leaves them out entirely.) The second fix is changing what is offered and how it's offered. The historian Jonathan D. Cohen, whose book Losing Big documents what legal sports betting did to a generation of young men, gave me a good list when I interviewed him about this topic last year. “I don’t think you should be able to bet on minor league British darts. I don’t think you should be able to bet as much as you want without ever having to provide proof of income and provide proof of how much money you have.” The prediction markets should be more mindful about the markets they offer, impose some sort of waiting period on deposits, and run an affordability check before people stake serious money. Polymarket released a commercial starring LeBron James and Eli Manning in which they relabel sports betting as “sports trading” and betting as “buying.” More and more money is going to go into advertising these products as something that should be a part of an investment portfolio. But the math behind it doesn’t hold up. Markets FTW One of the best-performing food companies in a year in which AI has made all the headlines is an almost 130-year-old jam maker from Orrville, Ohio. It makes Folgers, Dunkin coffee, Jif peanut butter, Smucker’s jellies and jams, Hostess treats, Milk-Bone, and Meow Mix, but its biggest growth story is a frozen peanut butter sandwich with the crust off. J.M. Smucker reported last month that Uncrustables grew sales 12 percent from a year earlier. The share of American households buying them has continued to increase–household penetration currently sits at 27 percent, according to their recent earnings report–and the brand passed $1 billion in annual sales in fiscal year 2026. J.M. Smucker has made several innovation decisions for its Uncrustables, including a peanut butter and raspberry spread alongside higher-protein Uncrustables (no one can escape the protein craze). The company also collected $115 million in tariff refunds, which it’s putting into marketing and paying down debt. The stock is up almost 40 percent this year. In a world where everyone is betting on AI, a crustless sandwich might be a good hedge. Chart of the Week The share of Americans working or looking for work fell 0.9 percentage points to 61.6 between December 2025 and June 2026. Outside the pandemic, it’s the lowest number since 1976, and it’s one of the main reasons why unemployment is hovering around 4 percent. Fewer people looking for jobs means that fewer people count as unemployed, yielding a lower unemployment rate. This is usually interpreted as people giving up on the labor market. Alexander Bick at the St. Louis Federal Reserve broke the 0.9 into parts, finding that about 0.35 comes from a bookkeeping change with updated population estimates. Fewer working-age immigrants and more people older than 65 mean the survey gets reweighted and the participation rate falls. Another 0.14 points is aging, which takes about a quarter point off every year. That leaves 0.33 points of people leaving the labor force, and nearly all of it happened in June. The August jobs report showed labor force participation bouncing back to 61.6 percent from a fall to 61.4 percent in July, with 683,000 people reentering the labor force. But participation is still 0.5 points below January, and the trends–an aging population and strict immigration policy–point to this problem getting worse before it gets better. What I’m Watching September 11: August consumer price index (CPI). Inflation is elevated due to the energy shock from the Iran war. It has cooled off for two straight months, and a third month of cooling would take real pressure off the Federal Reserve. September 15 and 16: The Federal Reserve meets! The big question it faces is whether to raise rates or not. The market is increasingly pricing in a hike. President Trump is actively threatening tariffs if the Fed does not cut rates. September 16: August retail sales are out in the morning. September 30: Micron reports earnings. It makes memory chips that go into AI servers and its stock is one of the most successful this year. It’s a good proxy for AI demand. Disclaimer: The opinions expressed above do not necessarily reflect those of the presenting sponsor.