Building a CFTC-Regulated Event Exchange: Kalshi, Market Design, and Event Risk

An event exchange turns a precisely defined future outcome into a tradable payoff. Building one requires much more than a forecasting website: contract language, order matching, collateral, settlement, surveillance, liquidity, and legal authority must work together.

This lecture uses Kalshi to examine that design problem. It expands the supplied outline, replacing numbered reference fragments without a bibliography with dated primary sources. The outline’s participant percentages, institutional names, and historical aspirations are distinguished from verified facts. Legal milestones and product-document checks are dated as of 16 October 2026; this is a lecture on market design rather than an exhaustive account of every ongoing proceeding.

Learning goals

Explain binary event-contract payoffs and their price interpretation.
Compare direct event hedging with proxy hedges and understand residual basis risk.
Read a central limit order book and calculate spread, fee, and depth costs.
Distinguish exchange designation, contract listing, and court proceedings.
Separate fully funded binary positions from margined or leveraged products.
Evaluate liquidity, revenue, distribution, and market-integrity claims critically.

For background, see Financial Markets and Bond Mathematics and Counterparty Risk Optimization.

1. Why trade an event directly?

The supplied lecture uses Brexit as a motivation: an institution exposed to a major political outcome might otherwise trade currencies, equities, or option structures as proxies. Such instruments can respond to many other shocks, so their payoff may not track the event-related business loss closely.

An event contract instead specifies a direct payment based on a resolution rule. Examples could reference a published inflation threshold, an official policy-rate decision, or a weather measurement. Directness can reduce one kind of basis risk, but it does not remove differences between the contract’s definition and the participant’s actual exposure.

Hedge type	What it directly references	Main mismatch
Currency or equity proxy	A market price	Other economic forces can dominate the event effect
Option spread	A range or shape of market-price outcomes	The event need not produce the expected price move
Binary event contract	One specified event resolution	Actual business losses may vary in size or depend on additional conditions

Not every economically or culturally interesting question is suitable for exchange listing. A workable market needs a permitted product, objectively resolvable terms, a credible source, sufficient demand, and manageable integrity concerns. “Trade any event” describes an ambition, not a universal legal or operational capability.

2. Binary payoffs and probability-like prices

Let I_E equal one if event E resolves Yes and zero otherwise. A standard one-dollar Yes contract pays I_E; its complementary No contract pays 1−I_E. The pair pays one dollar in total under the same final binary settlement rule.

If a trader buys n Yes contracts at price c each, holds to final settlement, and pays total fees F, profit is:

PnL=n(I_E−c)−F.

At c=0.40 for n=100, the cost is 40 before fees. A Yes result pays 100, producing a 60 profit; a No result pays zero, producing a 40 loss. The maximum gain, maximum loss, and probability of either outcome are separate quantities.

When price resembles a probability

With a subjective probability p, expected profit before fees is n(p−c). A price of 40 cents can be described informally as a market-implied 40% likelihood, but it is not necessarily the true physical probability.

In an idealized pricing model with deterministic discount factor D, price=D×Q(E) under an appropriate pricing measure Q. In practice, risk preferences, hedging demand, collateral remuneration, constraints, fees, and thin order books can separate traded prices from a calibrated forecast probability. A last trade can also be stale.

Buying at 40 cents with 1 cent of fee per contract requires a subjective probability above 41% for positive expected profit under that simple hold-to-settlement calculation. Miscalibrated confidence can overwhelm a small apparent edge.

3. A direct hedge can be useful without being a prediction bet

Suppose a business loses 10,000 if a specified event occurs and loses nothing from that event otherwise. Assume an exactly matching, sufficiently liquid Yes contract costs 0.30, with no fees, financing costs, or settlement mismatch.

Buying 10,000 contracts costs 3,000. The combined event-related result is:

Outcome	Business event effect	Contract net profit	Combined effect
Event occurs	−10,000	+7,000	−3,000
Event does not occur	0	−3,000	−3,000

The hedge converts uncertainty into a fixed cost in this ideal example. It need not be an attractive standalone speculation to be useful to the business. Actual exposure, executable size, contract limits, timing, and fees can prevent exact replication.

For a random loss L and per-contract hedging payoff Y, an unconstrained minimum-variance hedge amount in L−nY is n=Cov(L,Y)/Var(Y)* when moments exist and Var(Y)>0. Limits, cash requirements, and uncertainty about that covariance modify the decision. A binary contract cannot perfectly hedge every continuously varying loss.

4. Central limit order books and the role of the exchange

A central limit order book (CLOB) collects resting bids and offers, matching compatible orders under the venue’s rules. Kalshi’s order-book explanation describes buying and selling through displayed price levels.

The structure differs from a bookmaker that directly offers customers a quoted wager from its own risk book. However, an exchange still sets listing and matching rules, charges fees, and governs access and settlement. Market makers can be counterparties and quote prices to earn spreads. “No bookmaker” does not mean “no conflicts, costs, or professional intermediaries.”

Reading a quote

Suppose the Yes bid is 0.48 and the Yes ask is 0.52. The quoted spread is 4 cents. Buying 100 contracts at the ask and immediately selling at the unchanged bid loses 4, before fees. The midpoint 0.50 is not necessarily an executable price.

Depth matters too. If 100 contracts are offered at 0.52 and the next 200 at 0.55, buying all 300 costs 162, for an average price of 0.54. A screenshot of the best offer does not guarantee that price for a large trade.

A limit order caps an acceptable price but may never fill. An immediately executable order can consume multiple levels. Cancellation, queue position, latency, and news arrival affect execution quality. A one-cent spread in one active contract is not evidence of one-cent spreads or deep capacity across the whole exchange.

Complementary sides

For an ordinary binary pair, a willingness to buy No at 0.60 corresponds economically to offering the complementary Yes exposure at 0.40. Bid/ask spreads and actual execution still matter; do not add two displayed asks and assume they must equal one exactly. The certain terminal payoff of a matched Yes–No pair is not a guarantee that any proposed two-leg trade is profitable after prices, fees, and execution risk.

5. “Natural risk” versus gambling: an economic distinction, not a complete legal test

The lecture contrasts pre-existing exposures—such as weather or inflation risk—with risk created solely by entering an entertainment wager. This helps explain a hedger’s motivation. The same event contract can nevertheless be used by a commercial hedger, an informed speculator, or someone seeking entertainment.

An exchange format and a real-world event do not settle every legal or policy classification. Sports and political outcomes already exist independently of the contract, yet whether related contracts are permitted or governed by particular laws can still be contested. Traditional betting exchanges can also use peer-to-peer matching.

Evaluate the payoff, participant purpose, contract design, and applicable law separately. The relevant market-design questions include information aggregation, genuine hedging demand, consumer understanding, manipulation incentives, and how disputes are resolved.

6. The regulatory breakthrough: designation is one milestone

The CFTC announced KalshiEX’s designation as a Designated Contract Market (DCM) on 4 November 2020. The CFTC release describes an order issued under the Commodity Exchange Act and the requirements applicable to DCMs. The exchange’s company account describes its position as the first federally regulated exchange dedicated to event contracts.

That narrowly framed historical description is more precise than claiming that no US event market or permitted prediction activity existed before Kalshi. Earlier markets and differing regulatory arrangements existed. Nor does DCM status mean a regulator has endorsed a contract’s expected return or guaranteed a participant’s capital.

Separate three questions:

Is the entity authorized to operate as an exchange?
Can a particular contract be listed under the applicable product rules?
Can a particular participant access that contract in the relevant jurisdiction and distribution channel?

An answer to the first does not automatically decide the other two.

7. Election contracts: understand the procedural history

On 22 September 2023, the CFTC disapproved KalshiEX’s congressional-control contracts, following the exchange’s self-certification submission. Kalshi challenged that decision in federal court under the Administrative Procedure Act.

In September 2024, the district court vacated the disapproval. An administrative stay then interrupted trading while the CFTC sought a stay pending appeal. On 2 October 2024, the D.C. Circuit denied that stay request and dissolved its administrative stay, finding the required showing of irreparable harm lacking at that stage.

That appellate ruling allowed trading to resume while the appeal was pending. It was not a Supreme Court decision or a blanket final appellate ruling that every election-related contract was lawful under every legal theory. The opinion itself discusses earlier election prediction markets, so “first legal election trading in a century” should be understood as a promotional historical framing requiring a narrow definition, not a comprehensive statement that all earlier activity was illegal.

Later disputes are separate questions

The 28 August 2026 Ninth Circuit opinion in KalshiEX v. Assad addressed preliminary relief in Nevada, finding that Kalshi had not shown a likelihood that federal law preempted state gaming regulation as applied to its sports contracts, and remanding part of the dispute concerning other contracts. The court’s opinion illustrates why a 2024 election-contract milestone cannot be treated as a complete resolution of later sports, state-law, or access disputes.

These are dated decisions, not a claim that no later order, appeal, or product-specific limitation exists. For the lecture’s economic argument, the key lesson is that regulatory category, contract category, and procedural posture must remain distinct.

8. Collateral, mark-to-market, and settlement

For a simple fully funded binary pair priced at c and 1−c, the combined funded amount is one dollar, matching the combined terminal payout. A Yes buyer funds its purchase cost c; the complementary No exposure funds 1−c under this simplified structure. “100% collateralized” refers to covering the relevant maximum obligations, not requiring every 40-cent buyer to post an additional full dollar.

Kalshi’s historical event-contract explanation describes its fully cash-collateralized model. Under that model, buying at 0.40 creates a maximum purchase loss of 0.40 per contract before fees. Collateralization reduces default exposure but does not remove market losses, settlement delays, legal uncertainty, or operational risk.

Marking a position is not final settlement

If a purchased contract rises from 0.40 to a displayed market value of 0.65, the position has an indicative unrealized gain. Realizing a sale requires executable liquidity; final event settlement requires the resolution rules to be satisfied. A changing account valuation is not necessarily a continuously paid variation-margin cash flow.

The source agency’s publication, corrections, rounding, cutoff date, and dispute procedures can all matter. The market-settlement FAQ explains that determination can await official information. The event happening and withdrawable cash becoming available are not invariably the same instant.

Offsets and newer product types

The current collateral-return documentation describes specified event relationships where guaranteed payout combinations can reduce required locked collateral. This is rule-based offset recognition; it should not be assumed for any two correlated events.

Separately, current Kalshi documentation includes perpetual futures with isolated margin, described in How Margin Works. Those are different products with different leverage, settlement, and liquidation mechanics. The lecture’s statement that margin and leverage are only future possibilities is therefore not a timeless description of the entire platform. Always identify which product and agreement a margin claim concerns.

9. Market making: liquidity has costs and limits

Market makers post quotes, manage inventory, and hedge or offset exposures. Their spread revenue compensates for adverse selection, inventory risk, execution costs, and uncertainty around event information. A binary price can jump sharply toward zero or one after decisive news, so continuously quoting a tight spread is risky.

Kalshi named Susquehanna International Group (SIG) in a dated company liquidity announcement. The lecture’s additional reference to Citadel is not independently established here as a specific current Kalshi market-making relationship. Do not infer one firm’s role from another firm’s presence, or from unrelated brokerage order-routing disclosures.

The market-maker program description makes benefits conditional on meeting liquidity requirements and notes that market makers bear trading risk. Provider names alone do not measure liquidity. Examine spread, depth, execution size, uptime, and behavior around information releases.

10. Revenue and fees: calculate the actual trade economics

An exchange can earn transaction fees rather than betting against each customer’s outcome. Fees may differ by maker/taker status, market series, product, distribution channel, and scheduled changes. Revenue does not automatically equal profit or establish how much is reinvested in growth.

The event-contract fee schedule retrieved for this lesson is labeled effective 7 July 2026. Its general taker formula, before the specified rounding, is:

fee=0.07×M×C×P(1−P),

where C is contract count, P is dollar price, and M is the applicable series multiplier. Maker fees and exceptions are specified separately. Use the schedule’s actual rounding and series rules; a single flat cents-per-contract description can be misleading.

At M=1, buying 100 contracts at 0.50 gives a 1.75 fee under that example. The total outlay is 51.75, maximum net profit at Yes settlement is 48.25, and maximum loss is 51.75, assuming no other costs. The subjective break-even probability is 51.75%, not 50%.

An early sale can add another transaction fee as well as the spread cost. Brokers or other intermediaries can impose additional charges. A narrow advertised spread is not the same as low all-in cost.

11. Participants, brokerage distribution, and information quality

The supplied 30–40% institutional versus 60–70% retail split lacks a date, definition, and methodology. Share of accounts, contracts, dollar volume, open interest, and liquidity provision are different statistics. It is retained here as an unverified lecture snapshot rather than a current market-composition fact.

Brokerage integration is also a dated business development. Robinhood announced its prediction-market hub on 17 March 2025, initially offering contracts through KalshiEX, according to its own announcement. The outline’s mentions of Schwab and Fidelity should not be treated as verified integrations merely because they were named as ambitions. A distribution partnership also does not imply every customer or jurisdiction has access to every market.

Why diverse participation can help—and mislead

Commercial hedgers bring offsetting economic demands. Informed traders can incorporate research. Market makers facilitate execution, while retail access can broaden participation. But a market can still be thin, dominated by a few positions, or skewed by correlated beliefs and unequal information.

To evaluate prediction quality, compare time-stamped prices with later outcomes across many consistently defined contracts. Use calibration and proper scores, such as Brier score (1/N)Σ(pᵢ−yᵢ)², and account for fees, spreads, correlated events, and selection of visible markets. One correctly priced headline event does not establish general forecasting accuracy.

12. Building the exchange: the operational checklist behind the idea

Layer	Design question
Contract specification	What exact event, time zone, threshold, source, and revision rule determine payout?
Listing and access	Which product process, participant rules, and jurisdictional conditions apply?
Trading	How are orders matched, priced, canceled, and protected during outages or news?
Collateral and clearing	Which maximum obligations, offsets, and product-specific margin rules are enforced?
Resolution	Who determines outcomes, handles unavailable sources, and hears disputes?
Surveillance	How are manipulation, prohibited conduct, and conflicts monitored?
Liquidity	Can realistic order sizes trade before and after important information arrives?
Economics	Do fees support the venue while leaving useful hedging and trading opportunities?
Distribution	What do brokers and interfaces show users about costs, access, and settlement?

The best short contract headline can still conceal a poor resolution rule. A market linked to inflation, for example, must distinguish monthly from annual changes, adjusted from unadjusted data, a particular publication vintage, and the exact numerical threshold. An objectively worded question and a dependable resolution process are part of the financial instrument itself.

13. Practice questions

What are profit and loss before fees for 200 Yes contracts bought at 0.35 and held to a binary settlement?
Does a 60-cent quote establish a scientifically certain 60% probability?
At a 0.48 bid and 0.52 ask, what is the immediate round-trip loss for 100 contracts before fees?
A business loses 5,000 only if an exactly matching event occurs. How many one-dollar Yes contracts offset that event-state difference, ignoring costs and constraints?
Does DCM designation authorize every imaginable event contract without a separate listing or access question?
Was the October 2024 D.C. Circuit ruling described here a Supreme Court judgment resolving all event-market legal questions?
Does a one-dollar combined collateral requirement mean each 40-cent Yes buyer must independently post one dollar in addition to its purchase cost?
Why should a market-maker name or a narrow best quote not be treated as proof of unlimited liquidity?

Answers

Cost is 70. A Yes settlement gives 130 profit; a No settlement gives 70 loss, before fees.
No. It is a market price whose probability interpretation depends on pricing, participation, information, and trading frictions.
4, from 100×(0.52−0.48), before any transaction fees.
5,000 contracts under the stated exact-match assumptions. The purchase cost remains a fixed hedging expense.
No. Entity authorization, product listing, and participant access are distinct.
No. It denied a stay pending appeal and dissolved an administrative stay at that procedural stage.
No. The simplified pair is funded by c and 1−c across complementary exposures; check the actual product’s collateral rules.
Executable size, depth at subsequent prices, cancellations, adverse selection, and market conditions determine actual execution capacity.