Intro
Ian Deacon, a portfolio manager at Marathon Asset Management, makes the case for the listed US derivatives exchanges as a business the market has stopped paying for. He and Edward Chancellor work through the three worries that derated CME Group and Intercontinental Exchange, why the new entrant built to take CME's rates franchise has not moved anyone's book, and what happens when the exchanges point the same clearinghouse at GPU rental rates.
Guest: Ian Deacon, portfolio manager for US large-cap equities and global equities at Marathon Asset Management
Host: Edward Chancellor
Published: 28 August 2026 on The Capital Cycle Podcast
Episode page | 13 min
Key Takeaways
The exchange is the house, and it does not put up the chips
"the house doesn't even put up the chips as the clearinghouse is collateralised by the players themselves" — Deacon
Direction does not matter to the exchange, which is why margins sit above 60% on very little capital
Volatility is the product, not the hazard
"they all pay the same toll on the same day, and at most one of them can be right" — Deacon
The new entrant built to break CME has not moved anyone's book
FMX launched in 2024 with 10 of the world's largest banks behind it; two years on its share of rates futures is still in the low single digits
"A fee schedule can be copied overnight, but a liquidity pool can't"
Clearing is the second moat, and it works as a discount
ICE's margin methodology runs across more than a thousand energy contracts and lets a client net correlated positions
Exchanges can manufacture their own end markets
"They just need new things for people to be nervous about. And the world's rarely short of those" — Deacon
Regulation is the moat and therefore the risk
"this is the risk I take most seriously precisely because so much the moat is conferred by regulation" — Deacon
CME sued the CFTC in June; the CFTC has since proposed retightening the same public interest test it loosened
AI compute is starting to behave like a commodity, and CME is listing it
Renting an Nvidia H100 for a year cost about 40% more in March than in October last year
The stocks have lagged the index for a decade and now trade at or below the market
"CME trades at just over 20 times forward earnings. ICE somewhat below that, which is at or below the S&P market average" — Deacon
The Croupier's Take
Chancellor opens on Charlie Munger. He notes that "The late Charlie Munger liked to talk about how the finance sector resembled a casino, extracting layers of fees from clients, but always with the odds stacked in its favour" and asks Deacon to start with how the listed US derivatives exchanges extract what Munger called the croupier's take.
Deacon calls it a "Pretty apt analogy." In a casino "the players win or lose, and the house takes its cut of every hand either way", and a derivatives exchange runs the same way: buyers and sellers may win or lose, the exchange clears the trade, takes a few cents from each side, and does it again the next day.
The exchange has an advantage the casino does not. "the house doesn't even put up the chips as the clearinghouse is collateralised by the players themselves"
Why Volatility Is the Product
Volatility does the opposite thing to an exchange than it does to a bank. "for a bank or insurer, a volatility is the thing that endangers the balance sheet, whereas for exchanges, it's what makes people transact"
Deacon runs through who transacts: a pension fund hedging its interest-rate exposure, a refiner locking in the spread between crude and the products it makes, a macro fund taking the other side of either position. "they all pay the same toll on the same day, and at most one of them can be right"
Indifference to direction is what makes the economics extreme. Deacon says it has let CME Group and Intercontinental Exchange become "two of the most profitable large companies in America", businesses with margins above 60% on very little capital.
He discloses on air that both are Marathon portfolio holdings.
The Three Worries That Derated the Stocks
Both stocks have derated over the past year or so, which is why Deacon wanted to talk about them. He puts it down to three worries: new competitors, a laissez-faire regulator letting in lighter-weight rivals, and a sense that some of the growth might be speculative froth.
His framing of the question is a historical one. The issue is "whether the market spotted the beginning of the end of the exchange model or whether this is a rerun of 2021 in payments when everyone was convinced that the new fintech were about to displace Visa and Mastercard"
Marathon wrote about that episode at the time, and Deacon says the disruption has yet to materialize
Chancellor puts the worry in the show's own vocabulary: it would appear the capital cycle for the exchanges is entering a negative phase.
FMX and the Capital Cycle's Test Case
Deacon accepts the theory before he attacks the evidence. "the capital cycle would say that returns this high should pull in capital until those excess returns are competed away"
Sure enough, capital arrived. "FMX launched in 2024 with backing from 10 of the world's largest banks and trading firms, offering cheaper fees, a cross-margined deal targeting CME's interest rate franchise"
Two years on, the theory has not shown up in the numbers. "the share of rates futures is still in the low single digits and growth rates may look impressive in percentage terms, but that's what happens with small numbers"
The number Deacon actually watches is open interest — the stock of positions held at each venue — and that, he says, simply hasn't migrated.
A Fee Schedule Can Be Copied, a Liquidity Pool Cannot
Chancellor asks whether the incumbents are protected by network effects, and Deacon's answer is the episode's cleanest line. "A fee schedule can be copied overnight, but a liquidity pool can't."
What an exchange sells is not execution, which Deacon calls almost incidental. "It's the assurance that the other side of the trade will be there at a fair price today and in five years when you want out."
The moat compounds one contract at a time. "Every contract that stays put deepens that assurance."
A would-be disruptor therefore runs into the same wall the fintechs hit with the payment networks: a lack of distribution.
Clearing as a Price Cut No Entrant Can Match
Clearing is the second facet of the moat, and Deacon says it reinforces the liquidity benefit rather than sitting beside it.
ICE's latest margin methodology runs across more than a thousand energy contracts and lets a client net correlated positions — long gas here, short power there — and "post materially less collateral than the same book would need if it was scattered across venues"
Collateral is a real cost to the client, so the netting is a discount in disguise. "that's effectively a price cut that no entrant can match without the same breadth of product"
And it gets stronger as the pool fills. "Every new market participant improves the netting for everyone already in the pool."
Semi-Endogenous Growth, and the Real Options Detour
Chancellor asks Deacon to explain his own coinage, and does not pretend to like it: "I'm never sure what endogenous means, let alone semi-endogenous"
Deacon admits where he got it. "It's jargon that I misappropriated from macroeconomics textbooks" — from a conversation with colleagues about end-market growth 10 years ago, when the subject was video rather than exchanges.
Most companies are hostages to their end markets. A spirits company needs people to drink more, and that demand is generally treated as something the supplier cannot influence.
Exchanges are not in that position. "the exchanges on the other hand can manufacture their own end markets and effectively endogenise part of that growth" — the same clearinghouse, distribution and rulebook can be pointed at any new source of price volatility at almost no extra cost.
The new market is born with the incumbent already dominant. "because liquidity gravitates to where liquidity already lives, the new market's born with incumbent holding dominant share"
Deacon's summary of the model is the most quoted thing in the episode. "They don't need more oil burned, more bonds issued to grow. They just need new things for people to be nervous about. And the world's rarely short of those."
Chancellor's own detour is into real options theory. He recalls the dot-com boom, when people talked about the real options the tech companies held, says the likes of Google and Facebook did have them and did benefit from them, and adds that "that optionality isn't always priced in by the market effectively"
ICE's Mortgage Bet Is Waiting on the Housing Cycle
Chancellor presses on the case where semi-endogenous growth has not paid off yet: ICE went into the mortgage market, and so far it has not been profitable, which he attributes to the weak US housing market.
Deacon does not dispute the position, only the timing. ICE has made a number of acquisitions and assembled a portfolio of mortgage technology businesses that "have very strong, if not dominant positions, in both mortgage origination and servicing technology"
With interest rates where they are and no significant rebound in new issuance, he places the business closer to the trough end of its cycle.
The view is a hopeful one rather than a forecast. "we are hopeful that at some point that demand cycle will swing back up and there could be value or greater value in that part of the business"
Record Results, and Where the Growth Actually Comes From
Chancellor notes that the recent results are not reflecting any particular pressure, and Deacon confirms it with CME's first quarter. "in the first quarter, CME did 36 million contracts a day, a record of 22%. In fact, with records up in all six asset classes at once, which hadn't happened before."
ICE tells a similar story. "The exchange segment runs at around an 80% adjusted operating margin. Last quarter, group revenues grew 18%. Adjusted earnings grew about twice that."
Chancellor asks whether this is just the Persian Gulf, and Deacon says partly, no doubt — but "management points out that the buildup in open interest started well before this year's flare-ups", and he says the disclosed numbers back that.
He is careful to put it more conservatively than management does. Over the past decade, volume has grown much faster than the stock of open positions sitting behind it.
Energy is the sharpest version of that. "in energy, volumes are up over 50% on an open interest base that's actually shrunk"
The distinction matters because both sides of it pay. "it's not so much the stock of hedges that's grown, but the speed at which they're churned, and both earn fees"
The Regulator Is the Risk He Takes Most Seriously
Chancellor raises the Trump administration's laissez-faire approach to financial regulation as potentially the greatest threat to the exchanges, and Deacon agrees without hesitation. "this is the risk I take most seriously precisely because so much the moat is conferred by regulation"
The barrier to entry is a list of regulatory obligations, not a technology. To run a clearinghouse "you need designation, capital, default funds, and tolerance for being supervised as systemically important, which deters the casual entrants"
What the CFTC has done is let a competitor in through a smaller door. Kalshi, the prediction market, offers sports and political event contracts and lately Bitcoin perpetuals "under a much lighter wrapper than a traditional designated contract market carries"
Chancellor brings breaking news into the conversation: Kalshi has now said it wants to offer equity perpetuals. Deacon says these perpetual futures are more akin to contracts for difference, so it is not surprising to see it in equities, hitting the retail side.
None of it competes with the institutional franchises yet, and the word Deacon uses for the risk is drift. "The worry perhaps is drift that the lighter rulebook creeps towards rates, energy, equities, and the playing field tilts."
How the Exchanges Are Fighting Back
CME's response was litigation against the agency that supervises it. "CME sued its own regulator in June, which tells you that they take it seriously"
Deacon thinks the pendulum may already be swinging back. "Kalshi's defending something like 19 separate state and federal actions over whether its sports contracts are just unlicensed gambling"
In April the CFTC brought its first insider trading case involving event contracts. Deacon says that case involved a US service member who had allegedly used insider information to profit from the Trump administration's military operation in Venezuela, which led to the removal of President Maduro.
The regulator then moved against its own loosening. In June it "proposed rules retightening the public interest test, the same one had just loosened, which rather proves the point that a regulatory advantage or disadvantage can be withdrawn by the same body that granted it"
His base case splits the market in two. "our base case is that a lighter wrapper survives for small retail event contracts, but not for markets that are systemically important"
Selling Futures on the AI Boom Without Owning a GPU
The next manufactured end market is compute. Deacon says "AI compute starting to behave like a commodity" — prices that swing, buyers who need certainty, producers who want to lock in returns on enormous fixed investments.
The price move he cites is the trigger. "renting one of Nvidia's H100 chips for a year cost about 40% more in March than it did in October last year"
That behavior has a track record of creating a market. "that's exactly the sort of price behavior that summons a futures curve into existence. It happened with oil in the 70s and electricity in the 90s."
CME's move is the one it always makes. "What they've always done with a new source of volatility, they're listing it." GPU rental rate futures are coming with an index provider called Silicon Data, subject to approval.
ICE has announced its own compute contracts and is expanding its data center estate to sell low-latency access.
The punchline is that neither has to take a position in the technology. "they stand to benefit from the same AI boom without owning a single GPU and earn more, the less anyone can agree on what the compute's worth"
The Two Risks, and Why He Thinks They Are Priced In
Asked what could go wrong, Deacon names two real risks. The first is the regulatory one already covered: barriers built on regulation are only as durable as the regulator's convictions.
The second is pricing mix, and it is the quieter of the two. "If growth keeps skewing towards the cheapest contracts, at some point that starts showing up in revenue."
So he watches the per-unit economics rather than the headline. "the numbers I watch are fee per contract and open interest rather than the headline volume"
On valuation, the decade is the point. "ICE and CME have now lagged the index for a decade. CME trades at just over 20 times forward earnings. ICE somewhat below that, which is at or below the S&P market average."
Deacon's conclusion places CME among the businesses the AI trade has stranded. "CME now sits among the quality franchises left behind by a market preoccupied with the AI story, even while it builds the venues where the story will be hedged"
Chancellor closes on the pun the episode was built for: "the market's interest in the exchanges may have lapsed, but yours remains open", and Deacon agrees.
Deacon's bet is that the three things which derated the exchanges — a funded new entrant, a permissive regulator and a suspicion of froth — are all weaker than they look against a liquidity pool nobody can copy, and that a market fixated on AI is mispricing the two businesses quietly building the venues where AI risk will be traded.
Products, Companies & Tools Mentioned
CME Group (The centerpiece: a "record of 22%" first quarter on Deacon's account, records in all six asset classes at once, margins above 60%, and now suing its own regulator)
Intercontinental Exchange, ICE (The other holding: an exchange segment at "around an 80% adjusted operating margin", a margin methodology spanning more than a thousand energy contracts, a mortgage technology portfolio waiting on the housing cycle, and compute contracts on the way)
FMX (Launched in 2024 with 10 of the world's largest banks and trading firms behind it and cheaper fees aimed at CME's rates franchise; two years on, share is still in the low single digits)
Kalshi (The prediction market allowed in under a lighter wrapper — sports, political and Bitcoin perpetual contracts, now pushing into equity perpetuals, while defending "something like 19 separate state and federal actions")
The CFTC (The regulator that granted the lighter wrapper, brought the first event-contract insider trading case in April, and in June proposed retightening the public interest test it had just loosened)
Visa and Mastercard (Deacon's precedent: in 2021 everyone was convinced fintech would displace them, and the disruption has yet to materialize)
Nvidia's H100 (The rental rate that moved about 40% between October and March, which is why a compute futures market is being built)
Silicon Data (The index provider behind CME's GPU rental rate futures, subject to approval)
Marathon Asset Management (Deacon's and Chancellor's firm; CME and ICE are disclosed on air as portfolio holdings)
Google and Facebook (Chancellor's examples of dot-com era companies that really did hold real options and benefited from them)
Get the latest market chatter and takes as they happen:
X | Threads | Instagram | YouTube | TikTok | Facebook