Prediction Market Movers · Sep 23, 2026 · 30m 18m saved
With CJ Hetherington, Co-Founder and CEO of Limitless Exchange
Limitless Exchange began as an answer to a machine learning researcher who wanted a market in which swarms of software agents would be paid for finding out what is true on the internet, and would go broke if they were wrong.
Most new prediction market ventures in the US chase sports and elections. CJ Hetherington said US sports is already saturated, and is building a regulated, dollar-settled exchange for institutions that want to hedge risk, a side of the market he says has no incumbent.
"So I think building an exchange is hard."
CJ Hetherington, co-founder and CEO of Limitless Exchange, on Prediction Market Movers, founded the company in 2023 and launched it in 2024, when Kalshi and Polymarket had been live for several years. Limitless now runs an onchain exchange for customers outside the US, with much of that business in Asia-Pacific and Europe, and is going through the CFTC's review for a US exchange.
The full interview is covered here so you can skip it. 30 minutes of audio, 12 minutes of reading.
Here are the 12 arguments that matter.
Key Takeaways
No incumbent serves institutions in US prediction markets, while US sports is saturated
He said many people see Kalshi as a sports betting company
Limitless now aggregates liquidity from Kalshi and Polymarket, since neither will ever list the other's markets
User-created markets were harder than planned, because someone has to pay for liquidity before a price exists
Self-custody crypto products need different rules from exchanges that hold customer money
The developer who deploys an AI trading agent is responsible for it, as with the machine learning that Citadel and Susquehanna use
Margin products force the exchange and the broker apart, which fully collateralized event contracts do not
Customers have not asked Limitless for perpetual futures, and he does not see them as the core opportunity
Fewer new exchanges will launch, because bootstrapping liquidity takes a large upfront outlay
In Korea, Limitless runs a play-money market on League of Legends outcomes
Parlays shared by influencers could pay those influencers a share of trading fees through custom packs
1. Born From Agent Swarms
Chris Gerlach of Prediction News asked what lessons Hetherington took from Kalshi and Polymarket before Limitless launched. He started with how the company came to prediction markets at all: a machine learning researcher came to the founders looking for an economic system that could pull together swarms of autonomous agents to judge which information on the internet was true.
Agents that judged the truth well would earn money like a fund
They wanted this to work like a marketplace where if you were really good at determining what was true, then you would almost be like a profitable investment fund and you'd return capital to your LPs and all of that.
CJ Hetherington
Agents that judged it badly would run out of money
And if you sucked at it or you tried to manipulate the market as an agent or an agent swarm, you wouldn't be able to cover your computation cost. And so functionally, you would die.
CJ Hetherington
That sent the founders into prediction markets, which they saw as good infrastructure for that kind of tool.
2. Why 2023 Was Different
Prediction markets were not new to them, and volume had historically come and gone with election cycles. Hetherington gave two reasons the timing was better this time. Stablecoins let a platform offer financial services to the whole world from a public blockchain without anyone's permission. After the election, the regulatory climate changed, and getting a prediction market product regulated became relatively easier than it had been.
Kalshi, he said, was not doing much then. It had its lawsuit and did win the right to list election contracts, but Polymarket had the attention, and volumes were still low.
The founders expected the market to grow a great deal
But I think what we saw fundamentally is that this market is going to grow very large and there are a lot of opportunities today that haven't been executed on yet.
CJ Hetherington
3. User Markets Cost Money
The first product was what Hetherington called an Amazon of prediction markets. Users would create their own markets, share them with their audiences and earn part of the trading fees, which would give the platform a wide choice of markets and distribution led by creators.
The miss was the cost of setting a first price
I think what we underweighted is that when you're creating these markets you need to subsidize price discovery.
CJ Hetherington
Someone has to pay for liquidity before a market can arrive at a price for each outcome, he said, so the economics of user-generated markets were hard to get started.
4. Two Exchanges, Two Markets
Asked what Limitless offers that larger exchanges do not, Hetherington said most of the market is now focused on the US. Polymarket's volume is shifting heavily from its international platform to Polymarket US, which may mean many of its onchain users were in the US all along. He said he did not know. The onchain Limitless product targets consumers outside the US.
Inside the US, prediction markets have been aimed at retail customers, with a heavy focus on sports and elections.
The bigger US prize is hedging
But we do think there is a much larger kind of institutional risk transfer and hedging opportunity.
CJ Hetherington
Limitless is building a regulated US venue for institutional customers. Gerlach asked about institutional demand for an onchain product, and Hetherington said the institutional product will be cash settled in US dollars, much like an old school exchange, because institutions already have the infrastructure for that. It keeps some features of onchain trading.
Round-the-clock trading carries over
So we can have 24/7 trading and instant settlement and all this kind of stuff.
CJ Hetherington
5. Different Rules For DeFi
Gerlach asked how Hetherington's view of financial regulation had changed in three years.
He wants oversight
So, I believe there should always be, you know, regulation and oversight. I think that's very important in financial markets.
CJ Hetherington
Self-custodial DeFi products, where users hold their own funds on decentralized infrastructure, carry different risks from cash-settled products run by a custodian, he said, and should have different rules. They also raise questions about jurisdiction, because where a user sits matters little. He said he was excited that crypto companies and new kinds of derivatives can now go to the United States and get regulated.
On identity checks (KYC for people, KYB for businesses), he said the institutional product should have them. Onchain, identity checks have uses, such as cutting down on Sybil attacks, where one person poses as many to collect rewards. A platform that can prove a person is real would be more willing to pay them incentives. He still wants open infrastructure that anyone can use, provided they are not from a sanctioned country and have not committed a major financial crime.
6. Questions For Regulators
Asked what risks DeFi rules have to cover, Hetherington said it matters where funds come from and who a platform is dealing with, and he did not know what the perfect set of rules is. He expects clear frameworks and said regulators are getting there.
Both regulators are working on the new products
I think the SEC and the CFTC are working very hard on not only rules which govern traditional products but also you know kind of this new order or this new world.
CJ Hetherington
AI agents trading in these markets are another open question for regulators, he said. Most enforcement against crypto exchanges has concerned bank secrecy and money laundering, and the risk is different when the platform does not hold user funds.
Self-custody changes the risk
So, I do believe the risks are slightly different.
CJ Hetherington
7. Who Answers For AI Agents
Limitless allows AI agents to trade.
Agents already have tools to trade on Limitless
We recently launched our MCP.
CJ Hetherington
It also offers agent skills and an agent starter kit. Gerlach said he was looking forward to regulators trying to take an AI agent to court, and gave his own concern.
The host expects agents to become an excuse
I also think it will be a vector for abuse like people will probably hide behind AI agents and say oh like you know I didn't make this trade my AI agent made it for me
Chris Gerlach
He suggested regulators might prohibit black-box agents that cannot explain themselves and allow those that can be opened up. Hetherington put the responsibility on the developer. Citadel and Susquehanna have long used machine learning in their trading strategies, he said, and whoever deploys an agent should answer for it. He granted that this gets messier with self-replicating agents acting on their own.
Using AI does not change who is responsible
I do believe that you know if you are a participant of a regulated financial market shouldn't really matter if you use AI in your strategy or not but you should still be responsible for what you do.
CJ Hetherington
8. Perps Need A Broker
Gerlach asked why exchanges would struggle to offer both event contracts and perpetual futures, contracts with no expiry date. Hetherington said it depends on an exchange's stage, resources and management capacity, because each kind of exchange is very hard to build. Kalshi, he said, has done a good job with its perpetual futures rollout.
Two hard things at once are harder than one
And so, trying to do both of those things at once is naturally more difficult than just doing one of those things.
CJ Hetherington
An event contract is a fully collateralized swap, with every trade backed by US dollars. Futures are traded on margin, which brings more risk, more complexity and new systems. US prediction markets have produced exchanges that are also their own brokers, such as Kalshi and Polymarket. A margin product should generally be distributed through a futures commission merchant (FCM), a registered futures broker.
Margin splits the business in two
And so almost like in a margin product, you have mandated this exchange and broker kind of separation. You don't have the full vertical integration.
CJ Hetherington
A margin product is harder to get regulated, he said, and an exchange needs relationships with FCMs or its own FCM, which is a large task in itself. The product design is very different too.
Asked about demand from Limitless customers for perpetual futures:
Customers are not asking
Honestly, we haven't seen that much.
CJ Hetherington
Perpetual futures have been around a long time and can be traded in plenty of places, he said, and Limitless does not see them as its core opportunity.
9. Korea, Japan And China
Gerlach said much of Limitless's onchain business is split between Asia-Pacific and Europe, and Hetherington confirmed it. Every market is different, he said, including Japan, Korea, China and Australia within Asia-Pacific, so Limitless works out from first principles what each one needs.
In Korea, Limitless publishes research on the tax case
Like for example in Korea we operate a business called Limitless Research which is effectively like a forecasting desk where you know we publish a lot of research including research such as you know what would be the tax benefit to Korea of regulating prediction markets at home instead of leaking that flow to offshore products or to US exchanges.
CJ Hetherington
Limitless also launched a compliant, play-money forecasting product in Korea for League of Legends players, which he said is growing fast and lets young people trade on game outcomes without putting real money at stake. Gerlach contrasted Korea and Japan with China's strict anti-gambling policies.
Korea and Japan are strict too
Well, I mean, the anti-gambling policies are really strict in Korea and Japan as well.
CJ Hetherington
Limitless does not target mainland China, he said, though there are strong crypto communities in Chinese-speaking regions outside it.
10. No Institutional Incumbent
Gerlach asked what a new exchange can still offer in a crowded industry.
Liquidity is the hard part
So I think building an exchange is hard. it's mostly economically hard because you need to bootstrap liquidity, which comes at a great upfront cost.
CJ Hetherington
That takes a lot of capital, and he expects fewer and fewer new exchanges to launch for that reason. Limitless started early enough to build deep liquidity, he said. The market is large enough that incumbents miss opportunities, but he would not launch a US sports exchange.
US sports is full
So, I think US sports is very very saturated.
CJ Hetherington
In the US, he said, there is no incumbent on the institutional side.
How many people see Kalshi
I think there are a lot of people that view them as a sports betting company because you know 80 90% their volume comes from sports and parlays.
CJ Hetherington
New venues can also bring customer relationships, he said. Some firms that own those relationships may choose a brokerage approach on top of established exchanges, while others may build an exchange from the ground up and win share where incumbents do not pay attention.
11. Listing Rivals' Markets
Gerlach took customer relationships to be the advantage for a new exchange. Hetherington agreed, and said running a brokerage may be a better strategy than subsidizing all the liquidity up front. Limitless has just started aggregating liquidity from Kalshi and Polymarket on its own exchange. His reasoning: Kalshi will never list Polymarket's markets, Polymarket will never list Kalshi's, and the two will never work together.
Combining both gives more markets and more liquidity
then we'll have a richer market library than either of them could ever have and we'll have deeper liquidity.
CJ Hetherington
12. Room For Specialists
Gerlach asked about single-subject prediction markets, one for niche political questions and another for economics. Hetherington said some groups of contracts can become very large businesses on their own, and each needs its own focus, as each geography does.
Relationships could decide the niche
You might have an exchange that has a lot of relationships in the private credit world, for example, and they'll go and build an exchange primarily for private credit hedging.
CJ Hetherington
Any product has to start with one community and build it out until it sustains itself, he said, so specialization makes sense as a strategy.
Some niches may not stay small
But we might also find out that some of these product sets, some of these categories are so large that they become giant standalone businesses.
CJ Hetherington
Bonus Insights
Custom packs return to the user-market idea
Limitless recently launched custom packs, which let users build their own parlays, share them and earn trading fees from those who buy them. User-created markets had been hard, Hetherington said, because the cost of liquidity is often higher than the fees a little-traded market generates. Influencers, celebrities and tipsters already share parlays on social media.
Effectively, imagine that you could click that link, follow it, buy the pack, and then the revenue will go back to Drake.
CJ Hetherington
Friends can bet against each other during a game
Limitless also launched player-versus-player markets, where users trade against friends during a match on what happens next, such as a corner, a foul or a goal.
The US exchange is in CFTC review
And then for the US exchange, we're just going through the review process for the CFTC and standing the exchange up and getting ready to hopefully go live at some point in the near future.
CJ Hetherington
Gerlach said people he has spoken to who went through the CFTC process describe it as a lift.
X is the best place to follow Limitless
Hetherington pointed listeners to the company's X account, its Discord server and limitless.exchange, where he said the real-time probabilities are worth reading even for people who do not trade.
Hetherington's bottom line is that the open ground in prediction markets lies outside US sports: institutional hedging in the US and tailored products abroad, with Limitless listing Kalshi and Polymarket markets rather than paying to build all of its own liquidity.
Products, Companies & Tools Mentioned
Limitless Exchange (Hetherington's company: an onchain prediction market for customers outside the US, with a regulated US exchange in CFTC review)
Kalshi and Polymarket (The established US exchanges; Limitless now aggregates their liquidity and markets on its own exchange)
Citadel and Susquehanna (Trading firms he said have long used machine learning, his example for who answers for an AI agent)
League of Legends (The game behind Limitless's play-money forecasting product in Korea)
Stake (The betting site in his example of Drake sharing a parlay on Instagram)
Instagram (Where he said Drake shares his parlays)
Books & Resources Mentioned
Limitless Research (Limitless's forecasting desk in Korea, which publishes research on the tax benefit of regulating prediction markets at home)
Watch the full interview
If this was worth your time, send it to someone closer to the industry than you are.
Get the latest market chatter as it happens: