The Other Side of the Block

Strip the release to its mechanics and three roles fall out. Cantor Fitzgerald & Co. acts as an Introducing Broker: it arranges the trade for the client and earns a commission, but does not take the position. Kalshi is the Designated Contract Market whose block-trade rule (Rule 5.3, amended in a self-certified January 28, 2026 CFTC filing) allows two Eligible Contract Participants to agree a price away from the central order book at a price that is “fair and reasonable,” after which it clears like any other Kalshi contract, fully collateralized. The rule itself sets no numbers — it leaves Kalshi to “designate products” and “determine the minimum quantity thresholds,” and to specify the reporting window; the 25,000-contract floor and the fifteen-minute deadline come from the exchange’s help centre. A less-noticed clause limits block eligibility for commodity trading advisors to those registered or exempt, or foreign equivalents with more than $25 million under management. And Susquehanna Predictions, the event-contract unit of Susquehanna International Group, is the principal — the firm that quotes the price, takes the other side, and owns the risk from the moment of agreement until resolution or until it can find an offset.

That third role is where the economics live. A block is not a matched order; it is a bilateral risk transfer. Kalshi’s clearing removes the counterparty credit question, and the block framework removes the market-impact question for the client. What it does not remove is the inventory: somebody has to hold a $2 million binary position on a September rate cut or a $500,000 position on a Super Bowl winner until the event resolves. Cantor’s Pascal Bandelier framed the gap the product fills as institutions having “lacked the ability to transact at scale on a regulated exchange.” Susquehanna’s Joe Grubb framed it from the other side: “the next area of material growth for prediction markets will be large institutional risk transfer,” with the firm set up to “price and execute custom, tailored contracts for institutional counterparties desiring to hedge both general market and bespoke industry risk currently unserved by traditional insurance markets.”

Two things the release does not say matter as much as what it does. It does not say Susquehanna is the exclusive counterparty — Kalshi blocks are bilateral and any ECP can stand on the other side, so Cantor can in principle shop a block to Jump, DRW or Virtu. And it does not disclose the size of the thing: no block count, no notional, no fee schedule. The 25,000-contract minimum — which separate orders cannot be aggregated to reach — tells you the floor, and a $1 contract tells you a minimum block is a $25,000 maximum payout. Institutional size in this market is, for now, small by the standards of the institutions being invited in. Kalshi’s own first block, in April, was described as a “six-figure” bespoke carbon-auction contract with Jump Trading on the other side and Greenlight Commodities brokering for a Houston environmental fund.

The Cantor deal is not Susquehanna’s entry into prediction markets; it is the fourth distribution channel bolted onto a franchise that was built deliberately over three years, on top of two older businesses that turn out to be the relevant ones. The inventory below is what is publicly confirmed. Where a relationship is widely assumed but not confirmed, the table says so, because the distinction matters when you are assessing how many places the firm can lay risk off.

The competitive set

Susquehanna is the most visible maker but not the only serious one. Jump Trading has roughly twenty people on its dedicated prediction-markets team and took equity stakes in both Kalshi and Polymarket in exchange for liquidity — a fixed stake at Kalshi, a stake that grows with volume supplied at Polymarket US — and was the counterparty on Kalshi’s first block. DRW and Virtu are named by Kalshi as active makers; Wintermute quotes two-sided on Kalshi and Polymarket; Flow Traders and Akuna have desks; Kalshi itself runs an in-house unit, Kalshi Trading, which co-founder Luana Lopes Lara has said is “not profitable” and runs at a loss to support liquidity. Citadel Securities, IMC and Hudson River Trading were reported in April 2026 to have stayed out, citing regulatory uncertainty — though within the same week Citadel’s president said the firm was keeping an eye on the sector, and in July it hired Polymarket US’s former interim CTO. Citadel is still not a direct maker on either venue. Any of the active firms is an ECP and could be the other side of a Cantor block tomorrow. So the question is not whether Susquehanna has a monopoly on the pipe — it does not — but whether it has structural advantages in pricing and holding the risk that others would struggle to replicate. Sections 03 through 06 argue it does, and that those advantages are exactly the things that make the business dangerous.

A market maker that has just bought $2 million of risk from a Cantor client has, broadly, six things it can do with it. They are not equally available for every contract, and the whole craft of the business is knowing, before quoting, which of them will be open once the trade prints. Ranked by the cost of getting flat, they run roughly as follows.

Both examples are stylized. The sizes are chosen to be institutional by Kalshi standards and trivial by Cantor’s; the prices are illustrative round numbers, not quotes. What they show is the difference between a contract with a deep correlated market underneath it and one without — which is the difference between a brokerage business and an insurance business.

Example A — the FOMC block

A macro fund wants to own the September cut. Through Cantor it asks for 2,000,000 YES contracts on “Fed cuts rates at the September 2026 meeting,” a $2 million maximum payout. The Kalshi book shows 61¢ bid, 63¢ offered with a few thousand contracts of depth; the fund cannot get near its size on screen without moving the market. Susquehanna quotes 63¢ for the full block, one cent above mid, and the trade reports under Rule 5.3.

Susquehanna is now short 2,000,000 YES at 63¢: it collected $1.26 million in premium and owes $2 million if the Fed cuts. If it does nothing, it makes $1.26 million on no cut and loses $740,000 on a cut. The point is that it does not have to do nothing. A 25 bp cut moves the October fed funds future, which settles on the post-meeting rate, by roughly 25 bp times one minus the probability already priced — about 9.5 bp at a 62% market — and each basis point on a ZQ contract is worth $41.67. The quantity of futures that replicates the block is the notional divided by the dollar value of a full 25 bp move: 2,000,000 ÷ (25 × $41.67) ≈ 1,920 ZQ contracts, bought. On a cut the futures gain about $760,000; on no cut they lose about $1.24 million. Line the two up and the binary disappears.

This is the block business working as a brokerage business. The desk is paid a cent for providing size and immediacy, converts a binary into a rates position it runs every day anyway, and the client gets a fill it could not have had on screen. A competitor with a rates desk could do the same trade; the moat here is thin and consists mostly of being the firm Cantor called first.

Example B — the Super Bowl block

Now a sports-exposed business — a media company with a bonus clause, a retailer holding team merchandise, a sportsbook flattening its futures book — asks for 500,000 YES contracts on a team to win the Super Bowl, trading 12¢ bid, 13¢ offered in a thin book, six months before the game. Susquehanna quotes 14¢ for the block, two cents above mid, and is now short $500,000 of payout against $70,000 of premium, with a six-month holding period and no listed instrument that pays on the same event.

Walk the menu. Internalize: retail likes to own longshots, so retail flow in this contract is probably the same way as the client, not the other way. Kalshi book: the entire displayed size is a few thousand contracts; selling 500,000 into it would take the price to single digits. Cross-venue: Rothera, Polymarket US and Crypto.com list the same future; if one of them is trading 15¢, the desk can sell some there, but each venue’s book is as thin as Kalshi’s and the capital goes into a separate clearinghouse. Sportsbooks: Nellie’s accounts could place futures bets on the team at books showing, say, +650 (13¢), but US books price futures with 20–30% hold across the field, cap the stake, and will limit the account the moment it wins. Hedging $500,000 of payout that way would cost more than the $10,000 of skew the block earned. Correlated instruments: none — the team’s conference and division futures are the same risk in different wrappers.

This is the block business as an insurance business. The two cents of skew is not a dealing spread; it is a premium for underwriting a risk the desk will mostly hold. Whether 14¢ was the right price is a question that gets answered once, in February, which is why the desk’s edge in this example is its model, its capital, and its willingness to own the answer — and why a firm without those three things cannot compete for the trade no matter how good its connectivity is.

Sections 03 and 04 sort contracts on one axis — whether a correlated listed instrument exists. That is enough to separate a rate-cut block from a Super Bowl block. It is not enough to place an election, an earnings print, a merger close, or a military operation, because those contracts differ from sports and FOMC on a second axis that matters at least as much to the firm on the other side: how informed the counterparty is likely to be, and whether it can influence the outcome. The Fed leaks almost nothing and no trader can move the decision. A club can hedge its own relegation. A campaign has unpublished internals. An executive knows the quarter before the tape does. An active-duty soldier knew about the Maduro operation before Polymarket did — and, according to an April 2026 indictment charging five counts — three under the Commodity Exchange Act, one of wire fraud and one unlawful monetary transaction — staked about $33,000 and profited roughly $410,000.

Put the two axes together and the market has four quadrants, not two, and each quadrant calls for a different response from a block desk.

Politics: sports-like to hedge, worse on everything else

Election and policy contracts look like sports on the first axis — no correlated instrument, warehouse to resolution — and the piece’s Figure 3 treats them that way. Three things make them a class of their own for a block desk. Correlation: every House seat, Senate race and the presidency load on the same national swing, so a book that looks diversified across fifty contracts is one position, and a block in one race is a partial block in all of them. Tenor and capital: resolution is months or years out, every dollar is fully collateralized at the DCO, and there is no carry; the capital cost of warehousing a political block dwarfs a sports future that resolves in February. Who is on the other side: the informed counterparty is not a hedge fund with a model but a campaign with unpublished polling, a staffer who has seen the bill text, or — as Kalshi’s own enforcement record now shows — a candidate trading his own race (Minnesota state senator Matt Klein, a DFL candidate for the state’s 2nd Congressional District, took a five-year suspension and a $539.85 fine from Kalshi in April over a $50 bet on his own primary; two other federal candidates — one House, one Senate — were sanctioned alongside him). Kalshi has flagged more than 50 potential insider-trading cases and Polymarket has referred over 90 accounts to authorities so far in 2026 — different units, which the trade press has a habit of adding together; the CFTC’s Enforcement Division issued a prediction-markets advisory on February 25, 2026, resting insider-trading liability on CEA Section 6(c)(1) and Rule 180.1; and Senator Jeff Merkley said in June that the Commission’s proposed rule “fails to crack down on the explosion of gambling on prediction markets,” two months after he and six Senate colleagues wrote to the Chair demanding a rule that “prevents insider trading and corruption in the market.” A desk quoting a political block is underwriting all of that, with the skew it can charge capped by rule.

Corporate events: FOMC-like to hedge, worse than sports on information

The mirror image. A “Company X beats revenue” contract is close to a digital option on the stock’s reaction, and Susquehanna is one of the largest single-name options makers in the country; a merger-close contract maps onto the merger-arb spread the firm already trades. On the first axis these are the easiest blocks on the menu. On the second they are the most dangerous. The correlation between “beat” and “stock up” is imperfect (guidance, whisper numbers, the specific metric the contract names), so the hedge leaves a residual that is exactly the thing an insider knows. And the counterparty asking for a block two days before the print may be a fund, or may be someone with material non-public information — which brings securities law, not just the CEA, onto a CFTC venue. The jurisdictional question is live: on June 18, 2026 the SEC and CFTC issued a joint request for comment on derivatives definitions that asks, among other things, when an event contract on a single issuer “directly affects” its financial statements, financial condition or obligations and is therefore a security-based swap under SEC oversight. (The much-quoted illustration — whether a contract on a car maker’s quarterly vehicle deliveries is a swap or a security-based swap — is Davis Polk’s reading of the question, not language in the request itself.) Kalshi already lists company-specific KPI contracts (Tesla quarterly production, McDonald’s comparable sales); Polymarket lists weekly single-stock price contracts on names like Tesla, Nvidia and Apple. A block desk in this quadrant needs an information barrier, a surveillance posture, and a view on whether the contract it just filled will be re-characterised as a security before it resolves.

Two consequences for the Cantor channel. First, the contracts institutions are most likely to want — corporate events for equity funds, elections and policy for macro funds — are precisely the two high-asymmetry quadrants, which is where a block desk’s edge is thinnest and its legal exposure thickest. Second, the response is not only a wider skew. For the hedgeable-but-asymmetric quadrant it is an options-desk information barrier and a willingness to decline; for the unhedgeable-and-asymmetric quadrant it is a hard limit on size, a long-tenor capital charge, and a surveillance relationship with the venue. Which makes the block print feed doubly informative: it will show not just how large the institutional market is, but which quadrant it lives in.

Every element of the Cantor announcement that lives in a press release is replicable. Introducing-broker status on Kalshi is available to any FCM-connected broker; Interactive Brokers routes to Kalshi, CME and ForecastEx from one screen, Tradeweb has a minority stake and is embedding Kalshi data in rates and credit workflows, Talos shipped an RFQ and block interface in July, Greenlight Commodities brokered the first block. The block framework is Kalshi’s, not Susquehanna’s, and any ECP can be the other side. If the moat were distribution, there would be no moat. It is instead made of seven things that are hard to buy.

1 · A willingness to warehouse that has been tested

The Knicks night is the credential. Market makers collectively gave up about $22.4 million before fees on a game that traded $171.6 million, Susquehanna called it its biggest-ever sports loss without saying how much of that was its own, and the firm’s response was a one-line acknowledgment from its co-founder and continued quoting. Citadel Securities and Hudson River Trading — firms with deeper pockets — have so far chosen not to make markets directly. Capital is necessary for this business; the demonstrated appetite to lose some of it on a single binary, and a risk committee that has already signed off on that, is rarer.

2 · Two-sided retail flow on the two largest venues

Avenue 1 on the offset menu — internalizing a block against flow the desk is already handling — is available only to a firm that is quoting that flow. Susquehanna is the flagship maker on Kalshi, which cleared roughly $65.7 billion in Q2, and a co-owner and day-one provider on Rothera, which carries Robinhood’s retail flow — 22 billion-plus event contracts in the first half of 2026. A competitor for the Cantor block without that flow has to go straight to avenues 2 through 6, every one of which is more expensive. This is the same structural advantage SIG has held in listed options for decades, transplanted.

3 · Nine years of sports pricing

Roughly 86% of Kalshi’s Q2 notional was sports, 87% by June, and sports is the category with the fewest external hedges, which means sports is where the model is the hedge. Nellie Analytics has been pricing in-game wagers since 2017. Jump, DRW and Akuna are building sports teams now; the head start is measured in seasons of live-game data and in the sportsbook relationships that exist only because the accounts predate the limits.

4 · The options book underneath the financial contracts

Every “S&P above X on date Y” or “Bitcoin above Z” contract is a digital option, and SIG is one of the largest makers of the vanilla options that replicate it. For Example A-type contracts the rates hedge is available to anyone; for threshold contracts on indices and crypto, the ability to run the residual inside a vol book that is already delta-hedged all day is a genuine cost advantage.

5 · Structural entitlements

Designated market makers on Kalshi carry position-accountability levels ten times those of ordinary members and receive fee discounts, rebates and, the member agreement allows, revenue share. The published subsidy schedule is thin — half a cent per contract, $1,000 per market per day on the open tier, $50,000 per series per week for market makers — a rounding error for a block desk but a real constraint on anyone trying to build a passive market-making business from scratch; the market-maker program’s actual economics sit in a confidential appendix. And the bespoke-contract path Grubb described requires CFTC recertification for new terms — a process the desk has already been through.

6 · Ownership alignment

Equity-for-liquidity is now the market’s compensation model: Jump holds stakes in Kalshi and Polymarket; Susquehanna is a JV partner in Rothera (Robinhood controls it, MIAX kept 10%, SIG’s stake is undisclosed). A maker that owns the venue has reasons to quote through a bad night that a pure-fee maker does not, and venues know it. It also creates an obvious tension — the flagship maker on Kalshi co-owns Kalshi’s most credible competitor — which we return to under risks.

7 · Being first on the phone

Least durable, most immediate: Cantor’s release names one liquidity provider. Until a second maker wins a block through the channel, the flow defaults to Susquehanna, and block flow compounds — the desk that sees the most institutional tickets has the best read on where institutional risk is leaning.

The risk register below is organized by the question a risk manager would ask rather than by regulatory category. Several entries are specific to prediction markets and have no analogue in the options business Susquehanna grew up in; those are flagged.

The liquidity point, stated plainly

Prediction-market liquidity is deep exactly where institutions do not need it and thin exactly where they do. The Kalshi book is thick at 50¢ on a Sunday afternoon game and thin at 12¢ on a season future in August, at 8¢ on a relegation in March, and at every price on a bespoke contract. The block framework exists because of that asymmetry; Susquehanna’s job is to be the liquidity that the order book is not, and the price of being that liquidity is that the desk cannot itself rely on the order book to exit. That is why the offset menu matters more than the announcement: the size of the institutional block business Cantor can build is bounded not by Cantor’s client list but by how much un-hedgeable risk Susquehanna’s risk committee will let the desk carry into any given weekend.

The cleanest way to read the Cantor announcement is as the moment prediction markets acquired the structure every other derivatives market already has: a client, an agent, a principal, and a clearinghouse, with the principal’s balance sheet as the binding constraint. In rates and equities the principal can lay almost everything off and the business scales with flow. In event contracts, for most of what Kalshi lists, the principal cannot, and the business scales with capital and nerve. Susquehanna has more of both than anyone else in the category, which is why Cantor called it. It also has, for precisely that reason, the largest loss on record.

For institutions coming through the channel, the practical implications are three. Financial contracts will price tight because the other side can hedge; ask for size there. Sports and bespoke contracts will price wide because the other side is underwriting, not dealing; the block premium is the cost of the warehouse, and shopping it to a second maker will not close much of it until a second maker has built one. And the block print feed, not the press release, is where the size of this market will be revealed.

Kinetic Alpha has written before about what a guardrailed event-contract allocation might look like for retail, about the settlement plumbing underneath these venues, and about how settlement hardness decides which contracts can be attacked. This is the institutional counterpart: the market now has a dealer. Whether it has enough of one is a question for the next bad night.