Imagine a 30-minute interview with the CEO of OnlyFans where adult content is never mentioned. Given that it’s the focus of roughly 80 per cent of their video creators and is the main appeal of the site for its customers, you’d expect it to come up at least once.
It should seem equally strange, then, for Tarek Mansour, the CEO of the prediction market Kalshi, to give a wide-ranging interview and never talk about the thing driving more than 90 per cent of the activity on his site: sports betting.
In a talk this month at the Multicoin Summit in New York, Mansour discussed the long-term vision for his company, which allows people to trade shares of “event contracts”, priced based on the probabilities of various outcomes, and which recently raised $1bn from investors at an $11bn valuation:
Financialise everything. Everything can be a tradeable asset. The universe of what is tradeable today versus what it can be is so small . . . economic indicators, weather and climate, Covid, healthcare numbers, what Taylor Swift is going to do or not do . . . That can be applied to anything. Anything that has a difference in opinion can be traded someday.
One small omission from that list:
It was not the only time recently that Mansour avoided mentioning the thing driving ~all of his business . . . nor the only time that he showcased his keen awareness of a certain pop-cultural phenomenon:
Prediction markets came into the spotlight last year when they began offering Americans the opportunity to bet on elections en masse. This could be done legally on Kalshi, after a long legal fight, or via VPN on rival Polymarket, which only recently acquired approval from the CFTC to operate legally in the US.
But a far larger shift has taken place in the months after the election. Kalshi, equipped with its requisite Trump family member, has pushed past previously acknowledged limitations on futures contracts and transformed itself into a sports betting site.
Kalshi first dipped its toes into sports betting in the spring, but the volume has exploded since the start of the American football season in September. To date, there has been $16.8bn in trading volume on sports on the platform, against $4.9bn on all other topics.
Some regular-season NFL games have seen more than $60mn in trades. The top two of these games, alone, saw more volume than the New York City mayoral election, the most popular political event of the year on the platform.
Those “volume” numbers are somewhat fuzzy, as they include both sides of every transaction and accumulate totals for shares that are re-traded before the market resolves. An $0.80 share that is bought, sold and re-bought would count as $3 towards volume — including the $0.20 shares traded by another bettor as a counterparty — while it would only have counted as $0.80 towards the “handle” of a traditional sportsbook.
Because of their fee structure, their trading revenues appear to be even more lopsided than the volume, with approximately 89 per cent of all fees ever collected on Kalshi trades coming from sports. In recent months, that figure seems to hover closer to 95 per cent. Even for DraftKings, one of the two largest American sportsbooks, sports betting represented only 52 per cent of their revenue in Q3, against roughly 90 per cent for Kalshi in the same period.
There are two pieces driving the sports domination of Kalshi’s trading revenue:
The appeal to bettors and the very high volume this drives
A fee structure which allows Kalshi to extract more money per share on an average sports trade
Consider the limitations of election markets: they are relatively infrequent, often lopsided with low returns for backing the favourite, and many of the high-profile events occur simultaneously. Even when a market is “informationally” resolved, it can remain open for days or weeks before it is technically resolved, which locks up bettors’ money or forces them to cut into their winnings by cashing out early.
Sports markets, on the other hand, feature many high-volume, time-limited, scheduled events, spread out throughout the year. They resolve reliably within an hour or so of a pre-determined time and can be highly volatile, even through the last few minutes of trading, offering bettors many entry points with the potential for quick returns. There is also a large ecosystem of resources offering information to give bettors a real or perceived edge.
The incentives for most bettors to trade on a market just do not necessarily correlate with its research value or usefulness in the “truth machine”. So long as you think you have an edge, a dollar made off of an NBA match-up is the same as a dollar made off the presidential election, and you can probably get that dollar a lot more quickly.
Unsurprisingly, about 40 per cent of all shares traded on Kalshi’s markets are exchanged within two hours of the market closing, including more than half of all shares priced at 90¢ or more.
The other thing driving the profitability of sports markets for Kalshi is the fee structure. In part by chance, in part by design, they are extracting significantly more for sports trades.
The vast majority of Kalshi’s fees are collected on the taker side of trades. For most markets, they only charge fees on orders that are immediately matched — though nearly all of the markets where they do charge maker fees are sports markets.
Their taker fee formula has a quadratic relationship to the price of the shares being traded (that’s the one that forms a pleasant, arching curve without the tails), which encourages liquidity on high-priced, low-potential-return shares. These fees also round up to the nearest cent, slightly penalising smaller trades:
An analysis of trade-level data from Kalshi shows part of why this fee structure may yield higher revenues on sports markets. Relative to other markets, a larger proportion of sports market shares are traded in those middle price ranges, where the highest per-share fees are charged:
And though trade-level fee data is not publicly available, by doing some careful math on the trade data that is available, it’s possible to derive a rough approximation of the fees per share across categories, which tracks with what we could infer from that share price distribution:
A few caveats here.
First, the rules around Kalshi’s maker fee structure have changed several times over the past year. Most notably, their maker fees weren’t introduced until this spring, and they were restructured significantly over the summer.
Second, since May 2025, Kalshi has had a Market Maker Program in place, where a set of large market makers, such as Susquehanna Investment Group, provide liquidity on the platform and receive reduced fees.
Information about the identities of these market makers, the structure of their reduced fees, and which trades they are responsible for is mostly locked away in classified CFTC filings. We do know that, in addition to Susquehanna, the market makers include Kalshi Trading, an affiliated company that somewhat controversially trades on the platform. There are also some restrictions on the market maker trading hours and volumes — such as confining trading on sporting events to pre-game periods.
Very large traders also qualify for partial rebates on their fees, so the nominal fee at trade time may not represent real revenue. Like the market makers, there isn’t enough publicly available information to estimate the size or shape of these rebates.
In short, I’ve done my best, but there’s some level of uncertainty. Consider any maker fees here to be a rough “upper limit”, but even so, they’re a relatively small part of the picture.
So why the shyness around the meat of their actual business?
First, there’s the fact that the “sports-related event contract exchanges” they are facilitating may actually constitute illegal, unregulated sports gambling, in violation of the Wire Act, the Indian Gaming Regulatory Act and state-level regulators.
Kalshi has taken advantage of a somewhat “lax” regulatory environment under Trump’s CFTC, which regulates Kalshi as a “designated contract market”. The agency — down to one of five commissioners and still without a permanent chair — has been unwilling or uninterested in taking action against sports-related event contracts.
These were previously considered off-limits under a CFTC regulation which prohibits event contracts on “gaming” — as well as those related to terrorism, assassination or war — but only if the commission makes an explicit determination that a market is against the public interest. Kalshi argues that the lack of action makes their sports markets legal event contracts and that the CFTC’s authority to regulate those event contracts pre-empts state-level gaming laws.
This has led to many legal fights, some of which Kalshi is not winning. It turns out that states and tribes that regulate and tax gambling aren’t too happy about not getting a chance to regulate and tax something very ‘gambling-shaped’. Meanwhile, the states which don’t allow gambling are also not too happy about something very “gambling-shaped” advertising itself as “the first nationwide legal sports betting platform”. This latter group of states also notably includes Texas and California, and their 71mn residents.
Facilitating sports betting is also a less interesting story than being a “truth machine”. Kalshi talking about themselves as a sportsbook would be like WeWork talking about themselves as a commercial landlord.
Even if they have exploited a new “regulatory niche” here, it’s one that established gaming companies can, and are, jumping into. FanDuel and DraftKings, the two largest “traditional” American sportsbooks have each partnered with CFTC-licensed exchanges and plan to begin offering prediction market products.
Another well-resourced group that doesn’t see much of a distinction between prediction markets and gambling: the leagues themselves. They don’t seem particularly happy about an arrangement that removes a potential layer of regulatory oversight on platforms that may be susceptible to insider trading, at a time when integrity concerns are already under the microscope
If you try to place a bet on Kalshi on the Chicago Bears winning the Super Bowl (a boy can dream), you’d notice that there is no market for “Super Bowl winner”. Instead, there’s a market for “Pro Football Champion”, where “Chicago” is listed as an option, without a team name or logo. This is because the major sports leagues (with the notable exception of the National Hockey League) have not allowed prediction markets to use league data, logos or official designations.
In the spring, each of the three largest North American sports leagues — the NFL, NBA, and MLB — submitted letters to the CFTC expressing concerns about the ability of the agency to regulate gambling on these platforms. Last week, in a testimony submitted to the House Committee on Agriculture, the NFL’s vice-president for communications Jeff Miller was a bit more explicit:
Until such time that professional sports leagues and fans can be certain that effective game integrity and consumer protection measures can be enforced, sports-related events contracts should not be approved by the CFTC, and Congress should consider clarifying the definition of ‘gaming’ contracts in the prohibited categories of the Commodity Exchange Act. [h/t Dustin Gouker]
NCAA president Charlie Baker, who oversees college football, chimed in as well, warning that “prediction markets are not regulated at all . . . you’re basically talking about no rules, no oversight, no nothing. And that just feels catastrophic to me. Not just for us, but for everybody.” Major League Baseball sent a letter to its players as well, making clear that it considered prediction market trades to be gambling.
Suffice to say, it seems unlikely that the Chicago Bears logo will be appearing on Kalshi’s website anytime soon.
I’ll admit to (embarrassingly) being a bit of a sucker for the idea of the “truth machine” that these prediction market companies are trying to hold onto. Mansour wasn’t wrong when he recently described prediction markets as offering reporters “a tool, a stock, or a ticker or a market . . . a number now to all of these things that they’re talking about.”
As I described last year — in a piece that has admittedly not aged very well — the way that these markets move in response to news events can be informative, even if the specific prices aren’t necessarily “accurate”. This is particularly true for events that would be otherwise impossible to poll or model, such as Joe Biden dropping out of the 2024 presidential race, or who is likely to be the next Fed Chair.
It’s a compelling idea, even if the ethics of various markets can be pretty dubious. But for Kalshi, as it turns out, “truth machine” is more of a byproduct than a core business.