Crypto actual delivery: what exchange records must prove

A customer buys cryptocurrency on an exchange and sees the balance in an account. Has the exchange delivered the asset?

For an ordinary cash purchase, that question may sound academic. The answer matters when the exchange offered margin, financed the purchase, retained control of the asset, or later froze withdrawals. Under section 2(c)(2)(D) of the Commodity Exchange Act, some retail commodity transactions conducted on a leveraged, margined, or financed basis are regulated as if they were futures contracts unless an exception applies. One exception covers a sale that results in actual delivery within 28 days.

The phrase "actual delivery" turns a regulatory issue into an evidence problem. An account balance may prove what the exchange represented on its books. It may not prove that the customer obtained possession or control of the crypto asset.

The Commodity Futures Trading Commission put that distinction back on the agenda in October 2026. Its advanced notice of proposed rulemaking was issued on October 8 and published in the Federal Register on October 13. Comments are due December 14, 2026. The notice is an early rulemaking document. It requests public comment and states preliminary interpretations; it does not create a final crypto market regime.

Why the October proposal matters to disputes now

The proposed framework reaches farther than a simple question about whether a transfer appeared on a blockchain. It asks when an exchange's offer of leverage can bring a transaction within section 2(c)(2)(D), what possession and control mean for different token functions, and whether an internal book entry can amount to delivery.

The CFTC preliminarily reads the statutory word "offer" to include financing terms presented through onboarding documents, exchange terms, credit documents, or margin documents. Under that reading, a covered offer could attach to all transactions available through an account or product class. A customer who declines financing and pays in full could still have a "fully paid, open" transaction within the provision if the other statutory criteria are met. The agency has requested comment on that interpretation.

That makes document versioning important. A dispute may turn on the terms shown when the account opened, the products enabled for that account, whether margin was offered, and whether an exchange or connected party arranged the financing. The customer's final choice to pay in full would not answer every coverage question under the proposal.

The notice's preliminary view treats an exchange's internal ledger as constructive delivery, not actual delivery, while the transaction remains open. It states that actual delivery of a crypto asset involves a transfer from the exchange to the wallet address or account associated with the asset. The agency separately asks whether onchain trading protocols typically provide actual delivery because purchased assets generally arrive directly in a purchaser's wallet.

Those are preliminary agency views. They are useful now because they identify the records that will matter if the interpretation survives rulemaking. The possession, control, custody, and book-entry questions also track the facts examined in Monex.

Current law starts with meaningful possession or control

The main appellate authority is CFTC v. Monex Credit Co., a 2019 Ninth Circuit case involving leveraged precious metals rather than crypto assets. The court held that the actual delivery exception is an affirmative defense on which the commodity trader bears the burden of proof. It also held that actual delivery requires at least some meaningful degree of possession or control by the customer.

Monex did not hold that a buyer must always take personal physical possession. A third party depository can hold the commodity. The actual delivery defense did not support dismissal at the pleading stage because, accepting the CFTC's allegations as true, the metals remained in the broker's chosen depository, the broker retained exclusive control, the metals never changed hands, and the customers lacked substantial, noncontingent interests. The court stressed that a transfer of title alone does not establish actual delivery.

The CFTC's 2020 final interpretive guidance applied that reasoning to virtual currency used as a medium of exchange. It described two central features: the customer obtained possession and control of the entire quantity and could use it freely in commerce away from the execution venue, and the offeror and counterparty seller did not retain an interest, legal right, or control after 28 days. It also used a functional test that looked beyond contract language to how the transaction was marketed, managed, and performed.

The Commission withdrew that guidance in December 2025. The 2026 notice says current law on actual delivery is derived primarily from court decisions, then offers new preliminary interpretations for comment. Counsel should analyze the statute, controlling authority in the relevant jurisdiction, and the transaction's actual operation. The notice is a proposal, not current law.

A private key is evidence, not the whole answer

The CFTC preliminarily suggests that possession and control may require possession of the credentials, including private keys, associated with the relevant wallet address or account. That is stronger language than a general statement that the customer may request a withdrawal.

The notice does not say credential possession ends the inquiry. It says possession and control may require credentials and may depend on the asset's specific characteristics. Monex also recognizes that a buyer's agent can hold the commodity, so custody arrangements and retained control still require a fact-specific analysis.

The token itself can add another layer. The proposal says that possession and control may depend on the asset's characteristics. If a token conveys governance or staking rights, actual delivery may require unfettered access to those rights. The agency gives direct governance participation and staking without an intermediary's fee as examples, then asks for comment.

Counsel should separate four propositions:

1. The blockchain records a balance at an address.
2. A person or system can satisfy the signing conditions for that address.
3. The customer has a legal and operational right to direct use of the asset.
4. The customer can exercise the asset's relevant functions without an exchange retaining a veto, lien, forced sale right, or economic restriction.

One proposition does not automatically prove the others. Wallet control also does not identify the natural person who operated the wallet. Attribution still requires corroboration such as exchange records, device evidence, authentication logs, or an admissible signed message.

What an internal balance can and cannot prove

The October notice says crypto exchanges typically maintain customer property in omnibus accounts. When customer entitlements are recorded only on an internal ledger, the blockchain can show assets entering or leaving the exchange's custody perimeter. It generally cannot identify the customer credited on the private ledger, the terms governing that credit, or whether the customer could withdraw the full amount.

The October notice reflects this split. Its preliminary view treats an internal book entry as constructive delivery while the transaction remains open. It also asks whether a future crypto asset market should reconcile internal ledgers to public blockchain systems and asks for comment on proof of reserves for omnibus custody. The notice defines proof of reserves as a third-party audit attesting that reserve assets cover aggregate customer liabilities. It does not say such an attestation proves delivery to a particular customer.

Proof of reserves addresses an aggregate question: whether the intermediary's identified assets cover stated customer liabilities at a point in time under the method used. Actual delivery asks a transaction level question about the customer's possession and control. A venue can have sufficient aggregate reserves while restricting one customer's withdrawal. It can also complete a genuine withdrawal without publishing any reserve attestation.

This is the same architecture problem discussed in evidence in crypto perpetual futures disputes. Public transfers and private account records answer different questions. Counsel needs both when the claim crosses the boundary between the blockchain and the exchange's books.

Hypothetical: a paid purchase remains on the exchange

Hypothetical example: A retail customer opens an account with a centralized exchange. The onboarding terms make margin available for approved assets. The customer declines margin, pays the full purchase price for 20 units of Token X, and leaves the displayed balance on the exchange. Token X carries onchain governance and staking rights. Six months later, the exchange suspends withdrawals. The customer argues that the exchange had already delivered the tokens. The exchange points to its trade confirmation and internal account statement.

The trade confirmation shows that the exchange recorded a purchase. It does not establish actual delivery by itself. Counsel would need the versioned onboarding and margin terms, the customer's account eligibility, and evidence showing how the financing offer was presented. Those records address whether the fully paid transaction could fall within the CFTC's proposed covered offer analysis.

The custody evidence would include the exchange's wallet map, the address or omnibus wallet holding Token X, internal ledger entries, and reconciliation records tying the customer liability to onchain assets. Withdrawal policies and attempt logs would show when and under what conditions the customer could remove the full balance. Key management records would show who could sign, while access controls and administrative policies would show who could block a transfer.

Token function matters too. Staking records, governance delegation, reward entries, and fee schedules would show whether the customer could exercise Token X's rights directly or only through the exchange on terms set by the exchange. Any lien, security interest, clawback term, forced liquidation right, or account freeze authority would need separate analysis.

No single record decides the hypothetical. The conclusion depends on the applicable law and the functional relationship among the customer, exchange, custodian, wallet, and token.

The enforcement history explains why records matter

The CFTC has already used section 2(c)(2)(D) against crypto trading platforms. In a 2021 consent order, In the Matter of iFinex Inc., BFXNA Inc., and BFXWW Inc., the Commission found that Bitfinex offered off exchange leveraged, margined, or financed digital asset transactions to noneligible U.S. retail customers and operated as an unregistered futures commission merchant. The order described margin wallets, peer funding, account terms, customer location controls, forced liquidation, and the platform's role as counterparty.

The order was a settled administrative matter. The respondents consented without admitting or denying the findings, so counsel should not describe it as a judicial holding. Its value here is practical: regulatory analysis depended on platform records and operating facts, not a token symbol or transaction hash.

Records to preserve before the story changes

Counsel investigating a delivery dispute should preserve records by function. A generic request for an account history will miss the relationships the legal test examines.

Start with the account and offer record: every version of the onboarding flow, terms of service, margin agreement, credit terms, product eligibility, disclosures, customer acknowledgments, and interface screens shown during the relevant period. Request the data that establishes which version the customer received and when.

Then collect the transaction record: order, fill, trade confirmation, internal ledger postings, fees, timestamps, asset identifier, network, and account or subaccount. Preserve the venue's schema and field definitions so the export can be interpreted.

Map custody and control separately. Request deposit and withdrawal addresses, omnibus wallet mappings, wallet ownership records, signing policies, key custody arrangements, approval thresholds, transfer restrictions, freeze events, withdrawal attempts, rejection codes, and manual review notes. Obtain internal to onchain reconciliation records and any proof of reserves materials with the underlying method and snapshot time.

For functional tokens, add governance, delegation, staking, reward, slashing, and intermediary fee records. If the venue asserts a lien or right to liquidate, obtain the agreement, collateral records, margin calculations, notices, and each exercise of that right.

A blockchain evidence review should preserve raw chain data at identified blocks and document the provider, method, parameters, network, timestamps, and file hashes. The exchange production should be preserved in native form with audit fields and version history. The ConsensusIntel methodology explains why provenance and reproducibility matter when private records must be reconciled with public chain state.

Jurisdiction and limits

Section 2(c)(2)(D) is a federal commodity law provision. The threshold analysis concerns transactions entered into with or offered to persons who are not eligible contract participants or eligible commercial entities, on the leveraged, margined, or financed bases described in the statute. Different rules may govern institutional counterparties, securities, swaps, genuine cash market purchases, or transactions outside the United States.

The October notice expressly limits its preliminary actual delivery interpretation to section 2(c)(2)(D) and crypto asset transactions. It does not state how delivery works under federal securities law. It also does not decide ownership under state law, contractual entitlement, agency, bankruptcy priority, or whether a specific wallet belongs to a particular person.

Actual delivery is therefore one legal issue, not a substitute for the rest of the case. The evidence may show that an exchange recorded a liability, that assets existed in an omnibus wallet, or that a customer could initiate a withdrawal. Counsel must still connect those facts to the governing claim and the controlling jurisdiction.

Frequently Asked Questions

Q: Does an exchange account balance prove actual delivery?

A: Not by itself. It shows that the exchange recorded a balance for the customer. Actual delivery under the authorities discussed here turns on meaningful possession or control and the transaction's operation, not the ledger label alone.

Q: Must the customer personally hold the private key?

A: Monex does not create a personal key possession rule because it concerns precious metals, not crypto assets. The 2026 CFTC notice preliminarily suggests credentials may be required, while Monex recognizes that a third party can hold an asset for the buyer. The agency relationship, withdrawal rights, retained controls, and asset functions still need proof.

Q: Can a lien exist after actual delivery?

A: Possibly. Monex states that collateral status does not necessarily prevent buyer control. The 2026 notice asks for comment on onchain arrangements where an asset remains subject to a lien or liquidation mechanism but can still be used in commerce. The answer is fact specific, and the proposal is not a final rule.

Q: Does proof of reserves establish delivery to one customer?

A: No. A reserve review concerns aggregate assets and liabilities under a stated method and snapshot. It does not, without account level records, prove that a particular customer controlled a particular asset or could withdraw it.

If actual delivery affects jurisdiction, liability, or recovery in a crypto exchange dispute, contact ConsensusIntel before discovery closes. The first job is to identify which records can prove control and which records merely repeat the exchange's account balance.