Pricing the Future: Kalshi's Tarek Mansour on Winning the Right to Trade Elections

Mansour describes Kalshi as the first legal platform in the U.S. where people can bet yes or no on future questions. The instrument is neither a stock, a bond nor a commodity like oil. It is a contract on whether an event happens, such as rain tomorrow, a TikTok ban, Eric Adams being removed as New York mayor, or Donald Trump winning the next election.

To show how it works, host Sarah Guo makes a live trade on the election market while sharing her screen. She finds the price on Trump a little low at about 55%, goes all in with her $150 in cash, and the order fills immediately. Mansour points out that she can sell before the market closes, though she would pay the spread if she exited right away, and she would profit if his odds rose.

Mansour also highlights a feature Kalshi had just launched: money sitting in a position earns interest, because it is held in money markets and U.S. Treasuries. At the time he quotes 4.1%. He says this answers the common complaint that such bets are zero-sum, since you collect interest whether the position wins or loses.

Mansour says the company has held two convictions from the start. One is that prediction markets could be the next big thing in finance, because people like to take a view on things they care about, such as weather, climate or AI, instead of through loaded Wall Street instruments. The other is that he has never seen a financial market get big without being properly regulated.

He compares the approach to how Coinbase handled crypto: be legal, safe, compliant and trusted first. Before launching anything, Kalshi spent close to three years getting regulated by the CFTC, including roughly a year and a half in which he says he worked 14 to 16 hours a day writing the rules for how prediction markets could operate in the U.S. The company was approved and became the first to run prediction markets legally in the country.

The one thing the regulator did not want was election contracts, partly because a vocal minority in Congress believes they will break democracy. Kalshi sued, and Mansour says its board was initially alarmed at the idea of suing a regulator. He calls the fight brutal because the standard to win is far higher for a company than for the government. Kalshi won, and he says election trading is now legal in the U.S. for the first time in about a hundred years, this time regulated.

Mansour addresses two objections to election markets: that they are gambling, and that they could distort perceptions of the election through a reflexive loop. On the first, he draws on financial history. When grain futures were legalized in a 1905 Supreme Court decision, they were widely called gambling, even though they now look like the most boring hedging tool for farmers.

He separates two functions. Stock markets exist to allocate capital from those who have it to those who need it. Derivatives markets exist for risk transfer, moving risk from people who cannot bear it to people who can. Guo adds the origin story of the Dojima rice exchange in Japan, where farmers hedged price swings. Mansour adds that without futures, farmers face too much unpredictability, supply falls and bread prices rise.

On that logic, an election outcome is a risk that already exists, and nobody creates it for the purpose of betting. A climate-tech company is better off if Kamala Harris wins, and a crypto company is better off if Trump wins, so each has a real binary exposure to insure against. Mansour contrasts that with a roulette spin, which he calls recreation with no wider social benefit. He adds that markets also give price signals, and that a price equal to a probability is an even cleaner signal for, say, a TikTok creator wondering about a ban.

Mansour says that every market needs speculation. Without a mix of speculators, informed traders and hedgers, a market will be illiquid and have no activity. He also notes that people have gambled in some form since before writing existed, which he links loosely to evolutionary psychology.

He argues that life is full of risk-taking, from walking outside to starting a company, and that society tends to reward risk-takers. He cites Nassim Taleb's point that on risk, your grandmother can be a better guide than most data scientists, and adds his own view that much data science is hard to reproduce and shaped by publishing incentives.

He also compares casinos, which advertise the one person who won big on a slot machine, with startup culture, which celebrates Airbnb and Coinbase. The difference, he says, is that a company is net positive for society and a slot machine is not. His conclusion is that a life with risk in it is more fun, and he teases Guo that he thinks Taleb is conservative on technology while he himself is investing in AI.

At the time of recording, the election market had about $16 million in it, which includes Guo's $150. Mansour says Kalshi launched late because the lawsuit only just ended, and signups and volume had been growing roughly sixfold day over day for two days. He calls exchanges some of the hardest businesses to build because of the marketplace problem and the slow ramp of liquidity.

His plan is to start with retail traders, whose frequent, small trades build an initial layer of liquidity, and then draw in larger institutions that want to hedge things like cyber risk or a bill in Congress. He says the election is already seeing institutional flow, with demand for $10 million and $20 million trades and even up to $100 million. The constraint is whether institutions' compliance departments can onboard in the few weeks left.

On leverage, Mansour says it would create much more trading activity but would also make the system riskier. Kalshi now owns one of the handful of U.S. clearing houses, required for derivatives trading since the Dodd-Frank reforms after 2010, which centralize counterparty risk. He wants to let people take positions without posting all their cash within 12 to 18 months, but says it introduces credit risk and raises hard questions about how much leverage to allow a given user.

Asked how people should compare prediction markets with polls, Mansour says there is widespread confusion. A poll estimates a vote share, and if an unbiased poll showed Trump ahead by one point in Pennsylvania, that would point to Trump winning there. A prediction market instead prices odds. A 55% price for Trump does not mean he leads by five points in the polls.

He adds that markets move by more than the poll differences, because a small polling edge can translate into a much larger shift in probability. In his coin-flip analogy, the market describes a coin biased slightly toward Trump. It lands heads a bit more often, but tails can still come up. He thinks people misread this and need some education about odds.

Guo asks for a conditional trade, such as a bet on Trump that only counts if the Middle East is disrupted before the election. Mansour says conditional markets are the next big thing and his engineers want them too, probably within the next few months. They could answer questions such as what happens to GDP, inflation, crime or trade if a given candidate wins. He argues that markets are trustworthy because people like making money and dislike losing it.

Mansour says Kalshi is already a better gauge of the election than polling, and he calls 55% fair value for Trump while stressing that this does not mean Trump will win. He says Kalshi is unusual among prediction markets because it is American and regulated, so there is no foreign influence or tampering. It also has large institutional market makers, and he says it has no crypto or Trump bias.

He lists forecasts where he says Kalshi has done better than alternatives: inflation over the last two years, beating Bloomberg and the Economist survey, plus Fed decisions, TikTok, the Boeing CEO, the Omicron wave, and daily weather. His explanation is that an expert faces no consequence for being wrong, while a market pulls in many experts and amateurs and pays them to be right. This is the combination he calls wisdom of the crowd plus skin in the game.

Listing a new market now takes 24 hours, compared with 18 months for the first one and around two years for a new contract at a large exchange such as CME. About half of Kalshi's markets come from users through Market Builder. Mansour even imagines markets that forecast whether a movie will be good before release, which would stop producers from inflating the opening weekend before reviews come out.

Because Kalshi runs a regulated exchange, Mansour says he cannot trade on it and can only trade stocks. If he could, he would focus on year-long economic questions such as the Fed rate and inflation, and on whether tech CEOs will stay in their jobs. He says he thinks in terms of what is underpriced, not yes or no, and names Elon Musk leaving Tesla, priced around 2%, as one he would put closer to 15% to 20%, while calling it not financial advice.

He has also come to appreciate weather markets, since they resolve daily and let you improve through repetition. He has heard from users who scrape satellite data and model cyclical climate patterns at different frequencies. Some people, he says, make hundreds of thousands of dollars a month on weather alone.

Hedge funds, prop shops and Wall Street traders all use the platform, and he says some trading firms ask about recruiting from its leaderboards. Yet the best inflation forecaster over the past 18 to 24 months, by his account, is a man from Kansas who had never traded and just likes reading the news. Another team tracked how late the lights stay on at the Bureau of Labor Statistics to predict jobs data.