Risk & Market Structure

Custody risk in digital assets: what to establish

Custody is where a great deal of the real risk in digital assets sits, and where the questions an allocator asks are least standardised. This note sets out how custody differs in this asset class, the models in use, and the questions that separate a substantive answer from a reassuring one.

Manuel E. De Luque MuntanerCEO & Founder

21 August 202610 min read

Key takeaways

Digital assets are bearer instruments settled irreversibly, so custody failure is rarely recoverable — which changes what diligence has to establish before capital is committed.
Three models dominate — self-custody, venue custody and third-party qualified custody — and each moves risk somewhere rather than removing it.
“Segregation” means different things on-chain and in books and records; the distinction matters most in an insolvency, which is when it is discovered.
Proof of reserves without proof of liabilities establishes very little, and insurance is defined by its exclusions rather than its headline limit.
Trading strategies concentrate custody risk at venues, because collateral must sit where execution happens; the question is how much, for how long, and under what controls.

Why custody is a different question here

In traditional markets, custody is a mature, heavily intermediated function. Ownership is recorded in books maintained by regulated institutions, errors are usually correctable, and a chain of intermediaries provides both redundancy and recourse. Diligence tends to confirm that a familiar structure is in place.

Digital assets invert several of those assumptions. Control of an asset is control of a cryptographic key, transfers settle irreversibly, and a mistaken or fraudulent transfer generally cannot be reversed by an administrator. There is no equivalent of a trade break being unwound the following morning. The consequence is that custody moves from a hygiene check to a first-order investment question.

The three models, and what each one moves

Almost every arrangement is a variation on three approaches. None of them removes custody risk; each relocates it, and the relevant question is whether the relocation is deliberate and understood.

Self-custody — the manager controls the keys directly. Maximum control, and maximum concentration of operational and key-management risk in a single organisation.
Venue custody — assets are held at the exchange or broker where they are traded. Operationally convenient and often unavoidable for active strategies, but it merges custody risk with counterparty and credit risk in one entity.
Third-party qualified custody — a regulated custodian holds assets independently of the manager and the trading venue. This is the closest analogue to traditional practice, and it introduces its own questions about sub-custody, insurance and access.

What segregation actually means

Segregation is the term most often offered in response to a custody question, and it carries at least two distinct meanings that are easy to conflate.

On-chain segregation means client assets sit in wallets addressed to that client, verifiable independently. Segregation in books and records means the custodian's internal ledger attributes a share of a pooled, or omnibus, holding to each client. The second is entirely legitimate and widely used, but its protection depends on the custodian's records and on the insolvency law of its jurisdiction rather than on anything observable on a blockchain.

The distinction rarely matters while a custodian is solvent and functioning. It matters enormously in the one scenario diligence exists to prepare for, and that is precisely when it is too late to establish.

Questions that separate substance from reassurance

Most custody discussions produce confident answers. A smaller number produce evidence. These are the questions where the difference tends to appear.

Proof of reserves and proof of liabilities — an attestation that assets exist says nothing about what is owed against them. Reserves without liabilities is half a balance sheet.
Key generation and signing — how keys were generated, who holds the shares, how many are required to sign, and what happens if key personnel leave or are unavailable.
Insurance — the headline limit matters far less than what the policy excludes, whether it covers the specific loss types that actually occur, and whether the limit applies per client or across all of them.
Sub-custody — whether the custodian holds everything itself or delegates, and if so, to whom, under what liability, and with what visibility for the client.
Withdrawal controls — allowlisting, time delays, multi-party approval, and how an exception is authorised. Most published loss events involve authorised transfers, not broken cryptography.
Jurisdiction and legal structure — which insolvency regime applies, whether assets are bankruptcy-remote, and whether that has been tested or merely asserted.

Where custody risk shows up in a strategy

Custody is often discussed as though it were separable from investment strategy. In practice the strategy determines the exposure, because collateral has to sit where execution happens.

A strategy that trades actively across venues necessarily holds balances at those venues, and each balance is an unsecured exposure to that entity for as long as it sits there. The useful questions are therefore not whether venue exposure exists, but how large it is relative to the portfolio, how long assets remain there, whether balances are swept to independent custody on a defined cycle, and what limits apply per venue.

A strategy that is largely passive can hold assets with an independent custodian almost continuously, and its custody profile looks quite different as a result. Neither is inherently safer; they carry different risks, and the difference should be a deliberate choice rather than a by-product.

Custody risk is allocated, not eliminated

No arrangement removes custody risk. Self-custody concentrates it in the manager, venue custody concentrates it in the trading counterparty, and third-party custody transfers it to a regulated institution together with a dependency on that institution's controls, its sub-custodians and its jurisdiction.

What distinguishes an institutional approach is not the claim to have eliminated the risk, but the ability to say precisely where it sits, why that placement was chosen, what limits contain it, and what would have to change for the answer to change. A manager who can answer that has thought about the problem. A manager who says custody is fully covered has usually answered a different question.

Block Asset Management

How Block Asset Management helps

Custody sits inside our operational due diligence rather than alongside it. We assess it as an investment risk, at the level of each manager and each structure, and we treat an unclear answer as a finding rather than an inconvenience.

Custody assessed as a first-order risk

Custody arrangements are one of the assessment areas in our operational due diligence, evaluated before an allocation and revisited during it — not confirmed once at onboarding.

Venue and counterparty exposure examined

We look at where assets actually sit during the investment process, how long balances remain at trading venues, and what limits apply to each counterparty.

Evidence rather than assertion

We seek documentation, attestations and operational detail, and record where a manager can evidence a control and where the answer rests on assurance alone.

Structure-specific review

Our structures differ in their arrangements, including whether a depositary is appointed, so custody is assessed against the specific structure rather than described in the aggregate.

Conflicts examined, including our own

Relationships that could affect an assessment are identified and disclosed, on the principle that a conflicts review which excludes the reviewer is incomplete.

Ongoing monitoring

Custody arrangements change. Material changes at a manager or its custodian are treated as new information requiring review, not as an administrative update.

Custody is one of the few areas in digital assets where a failure is usually final. That asymmetry is what justifies the disproportionate attention it deserves relative to how briefly it is often discussed — and why an allocator is right to press past the first confident answer.

If your organisation is assessing how a manager custodies assets, our investor relations team can discuss how custody is examined within our operational due diligence process.

Important information

This material is provided for information purposes only and is intended for professional and qualified investors. It is general commentary on custody practices in digital asset markets and does not constitute investment, legal, tax or operational advice, nor an offer, solicitation or recommendation of any strategy or financial instrument. It does not describe the custody arrangements of any specific vehicle; those are set out in the relevant offering documentation. Digital assets are volatile and involve significant risk, including the possible loss of the entire amount invested. Due diligence reduces but cannot eliminate risk, including the risk of custodian or counterparty failure. Past performance is not a reliable indicator of future results.

Related research

Subscribe to our investor updates

Request Information Register for access