Custodial crypto lending, assessed as a counterparty question

The custodial model is simple to describe and that is most of its appeal: transfer assets to a firm, and the firm pays a yield. No collateral to manage, no liquidation to watch, no wallet to secure.

The simplicity is the product. It is also where the risk goes, because everything the user no longer has to think about is now something the firm does on their behalf, out of sight.

This assessment grades disclosure and structure, not desirability. We recommend nothing.

The structure

What the model involves

Asset custody	The firm holds the assets; the user holds a claim against the firm
Yield source	Onward lending, trading, or deployment — disclosure varies enormously
User's legal position	Frequently an unsecured creditor, depending on terms and jurisdiction
Rate	Advertised, variable, and set at the firm's discretion in most terms
Withdrawal	Subject to the firm's liquidity and to any terms permitting suspension
Segregation	Whether user assets are segregated from firm assets — a critical, variable term

The question that decides everything

Who pays the yield?

If a firm advertises a return, that return is generated somewhere. It may come from lending to borrowers who post collateral, which is a describable and relatively legible activity. It may come from lending to trading firms, which is a credit decision the user cannot inspect. It may come from the firm’s own trading. It may, in the sector’s documented history, have come from related parties or from new deposits.

A firm that explains its yield source specifically has told the user what risk they are taking. A firm that does not has told them nothing, and the advertised percentage is not a substitute.

What the better operators do

It would be unfair to assess this model purely by its failures, and some firms disclose considerably more than others.

The practices worth crediting: publishing the yield source in specific terms; stating clearly whether user assets are segregated; publishing proof-of-reserves attestations, with the significant caveat that a reserves attestation shows assets and not liabilities unless the exercise covers both; naming an underwriter and an insured event rather than using the word “insured” alone; and stating plainly what happens to user claims in an insolvency.

Firms doing all of that exist. They are not the majority, and the gap between the best and worst disclosure in this model is wider than in any other structure this site covers.

The failure mode, documented

The sector’s last cycle produced insolvency proceedings for several large firms operating this model, and those proceedings are matters of public record. The pattern in them was consistent enough to be instructive: users discovered the nature of their claim, and the identity of the counterparties their assets had funded, only after withdrawals were suspended.

We report that as history. We make no claim about the solvency of any firm currently operating, and readers should treat anyone who does — in either direction — with caution.

What we will not say

We will not state or imply that any custodial lending firm, or this market, is regulated, authorised, licensed or protected by any compensation scheme. Where a firm holds a specific registration, that registration covers specific activities and frequently not the lending arrangement itself. Readers should check what a permission actually covers rather than treating its existence as reassurance.

Pros and cons

The honest ledger

Pros

Genuinely simple to use — no collateral management, no liquidation exposure, no key custody
The better operators publish yield sources, segregation terms and insolvency treatment
Proof-of-reserves attestations are more common than they were, and are worth something
Fixed-term products at least make the commitment period explicit

Cons

The user is typically an unsecured creditor, a fact rarely stated as plainly as it should be
Yield source is frequently undisclosed, making the risk unassessable from outside
Withdrawal depends on the firm's liquidity and on terms that often permit suspension
Reserves attestations show assets, not liabilities, unless the exercise explicitly covers both
'Insured' is widely used without a named underwriter, insured event or insured party
Documented insolvencies in this model resulted in substantial depositor losses

Verdict

Our assessment is that this model is graded down not for what it does but for what it can decline to explain. Where a firm publishes its yield source, its segregation position and its insolvency treatment, a user can at least make an informed decision. Where it does not, the advertised rate is the only information on offer and it is the least informative number available. Total loss of deposited assets is a documented outcome in this model. Nothing here is financial advice.