Hey readers,
It's Dylan Matthews here, filling in for Kelsey Piper, who has been busy on some other stuff.
If you’re an American looking to make some money betting on future elections, I have some bad news.
The Commodities Futures Trading Commission, the federal agency tasked with regulating financial products like derivatives, has voted 3-2 for a proposal to ban “event contracts” on elections, but also on sports and on events like the Oscars. The rule targets prediction markets, sites like PredictIt or Kalshi that let you place real money on events happening in the future. It probably won’t be in effect until after November, but if you want to bet on the 2026 midterms, you may be out of luck.
The case for prediction markets is simple: They give observers valuable information about the future. The information may seem low stakes in the case of the Oscars or sports, but obviously who controls the presidency is of public interest, and with polling getting harder and harder, we need all the help we can get in forecasting and understanding election results.
I find these arguments pretty persuasive, and the arguments raised against legally allowing prediction markets frankly silly. Sports betting is now legal in 38 states and DC. It seems incredibly perverse that bets on the Knicks and Pacers would be legal but bets on Senate races that provide actually useful information to citizens the same way polls do would be banned.
At the same time, I’m skeptical that a bad legal regime is really what’s holding prediction markets back. Nick Whitaker and J. Zachary Mazlish have a smart essay in Works in Progress outlining a theory I find persuasive: Prediction markets aren't working because they don't provide enough value to the kind of people you need to make a market work.
Who puts money in a prediction market?
While the proposed CFTC ban is very broad, prediction markets on subjects other than elections are usually legal. At Kalshi right now, you can bet on what the Rotten Tomatoes score will be for Francis Ford Coppola’s comeback movie Megapolis (bets are currently hovering around 50), whether the Fed will cut interest rates before the end of July, and how thin Arctic Sea Ice will be next summer.
But these markets have not exactly taken the world by storm. Only 14 markets on Kalshi have $100,000 or more bet on them. That may seem like a lot, but compared to the stock market or sports betting it’s a pittance. What's more, the top four markets are all about Fed interest rates, which, as Whitaker and Mazlish note, you can already bet on through the much larger futures market. The novel opportunities prediction markets offer, like betting on Megapolis’s Rotten Tomatoes score, are less utilized.
In a world where the markets are efficient and reasonably well-used, there are strong theoretical reasons to think the prices they produce will be accurate. If they weren’t accurate, and it was possible to know that, then someone could be making a ton of money betting in a different way. And once they made that bet, the market would move and become more accurate.
For prediction markets to be obviously wrong, someone would need to be leaving easy money on the table, and that doesn’t normally happen in a capitalist society.
But when they aren’t well-utilized, this argument doesn’t follow. The price might be wrong simply because the amount of money at stake is too small for people who know better to bother wagering, because the amount they can win isn’t worth the trouble.
This is the heart of Whitaker and Mazlish’s case. They divide participants in betting markets into three types: savers, who try to grow their wealth; gamblers, for whom they’re entertainment; and “sharps,” who try to make money from understanding the market better than others.
For none of these groups are prediction markets very useful. You should absolutely not invest your 401(k) in a prediction market; whereas the total value of the stock market grows over time, prediction markets are zero-sum. If you take your savings out of the S&P 500 and put it in buying both “yes” and “no” on the “will The Tortured Poets Department top the Billboard charts for over 10 weeks” contract, you will absolutely lose money. Savers are out.
Gambling is a more plausible case for prediction markets. But Whitaker and Mazlish observe that in the UK, where this is all much less regulated, the popularity of sports betting completely swamps that of any other kind of contract. Yes, people like to gamble — but just about sports.
That makes sense: Sports happen in real time, where the odds are fluctuating constantly, and where betting in real time can give you a certain rush. In-game betting, for instance, is especially popular. Other sorts of questions prediction markets might help us understand — Who’s going to be the next president of Iran? Will China attack Taiwan? Will bird flu become a pandemic? — don’t have this dynamic. They aren’t exciting.
"Simply put," as Whitaker and Mizlash write, "most things that we might want to know about the future aren’t much fun to bet on."
That leaves the sharps (sharks?), who are trying to make money by being more right than the next guy. Prediction markets would be great for them … if there was anyone for them to bet against. But without savers and gamblers to profit off of, the gains for sharps are limited. And if everyone else investing is also a reasonably smart sharp, isn’t that a signal that they’re probably right, and you’ll probably lose betting against them?
Without much to offer sharps, savers, or gamblers, prediction markets are left with … no one.
There’s no harm in trying
The main legal prediction market in the US, Kalshi, is pretty small but its predictive record is still decent. More to the point, limits in the power of prediction markets aren’t a good reason to ban them, as the CFTC is attempting. In fact, it’s hard to find any good reason to ban them.
Six Democratic Senators wrote to the CFTC last year that “billionaires could expand their already outsized influence on politics by wagering extraordinary bets while simultaneously contributing to a specific candidate or party.” But billionaires are already able to place unlimited bets on stocks in industries like clean energy or firearms whose fortunes depend heavily on who’s in charge of the government; prediction markets would merely make the information driving those bets easier for the rest of us to access.
The idea that these billionaires could swing elections just to make money on bets is similarly far-fetched, as the writer Maxim Lott points out: “the thing with election manipulation is that even the most powerful individuals are rarely in a position to tip an election. It’s much harder to flip an election than a sports match, because of the number of people involved.”
More empirically, Britain has had a tradition of electoral betting dating back to before the Magna Carta and has had a legal market since 1961, without any of the horror stories the senators invoke coming to pass. They’ve done a decent job of predicting election winners, and you won’t find any wild stories about how Tony Blair won in 1997 because Lord Sainsbury really wanted to make sure his “Labour wins” contracts cashed out.
But I agree with Whitaker and Mazlish that real-money prediction markets need a better value proposition to succeed, even with more reasonable regulations.
I’ve been surprised at the vibrancy of Manifold, a prediction market that only uses play money. The whole point of prediction markets is that they make you have “skin the game,” something to lose if you’re wrong. All you lose in Manifold is “mana,” a fake currency.
But the frivolity might be part of the point. Precisely because you don’t put real money up, it’s easier for people to have fun making silly markets and betting on silly stuff. It’s a free form of social media engagement, like arguing on Twitter or Reddit.
Maybe the more important thing is being simple and fun.
—Dylan Matthews, senior correspondent
