What Actually Backs a Stock Token: Reading the Reserve Disclosure Instead of the Marketing

The claim, and what it leaves out

A tokenized equity is sold on one sentence: each token is backed one-to-one by a real share held with a custodian. That sentence is usually true and almost never sufficient. It describes a quantity, not a legal position. Two products can both hold exactly one share per token and still differ in whether you own that share, whether a liquidator can reach it, who is allowed to move it, and how quickly anyone outside the issuer can tell that the share is still there. The interesting work is in reading the disclosure that sits under the sentence.

Four questions that separate the structures

The first is what the token is a claim on. It can be a direct beneficial interest in an identified share, a share in a pooled vehicle that holds the shares, or a contractual obligation of the issuer whose performance happens to be hedged by holding shares. The third is not fraud, it is a note, and it behaves like a note if the issuer fails. The second question is where the shares sit: with an independent regulated custodian, with an affiliate of the issuer, or in a brokerage account that the issuer itself controls. The third is whether the holding is segregated and titled for the benefit of token holders, or commingled with the issuer's own assets. The fourth is what the issuer is permitted to do with the shares while it holds them, specifically whether they can be lent, rehypothecated, or posted as margin.

Segregation is the load-bearing word

Segregation is the difference between a custody failure and a credit loss. If the shares are held in a segregated account titled for token holders, an insolvency of the issuer is a mess of process: someone has to be appointed, the account has to be identified, the register of token holders has to be reconciled against it, and the shares eventually get distributed or sold. If the shares are commingled on the issuer's own books, token holders are unsecured creditors standing in line with everyone else, and the fact that the quantity matched perfectly on the day before the failure buys them nothing. This is the same distinction that governs client money rules in conventional brokerage, and it does not become less important because the claim ticket is a token.

What an attestation is, and what it is not

Most reserve disclosures are agreed-upon procedures engagements, not audits. The distinction is precise and worth carrying. In an agreed-upon procedures report, the issuer specifies what the accountant should check, the accountant checks exactly that and reports the findings without an opinion, and the report typically says so in its first paragraph. An audit produces an opinion on financial statements under a standard the auditor did not negotiate with the client. Neither is worthless, but an attestation that confirms a share count at a point in time tells you nothing about the other 364 days, nothing about whether the shares were encumbered, and nothing about liabilities the issuer holds elsewhere that could reach the same assets.

Reading the three dates

Three dates govern how much an attestation is worth. The as-of date is the moment the count was taken. The publication date is when anyone outside the issuer could see it. The gap between them is the window in which the disclosure was true of a past that has already been replaced. A monthly attestation published three weeks in arrears is describing a position up to seven weeks stale by the time the next one lands. Point-in-time counts also invite window dressing, the practice of borrowing assets to be in position on the measurement date. The defence against that is not a bigger accounting firm, it is unpredictable timing or continuous proof.

The onchain half of the problem

Backing has two sides and the reserve is only one of them. The other is the liability: the number of tokens in existence. Proof of reserve that verifies the asset side while the token contract can mint without constraint proves half a ratio. What matters is whether the supply is programmatically bounded, whether minting is controlled by a multisignature arrangement or a single key, whether the contract is upgradeable and who can upgrade it, and whether freeze and seize functions exist. Freeze functions usually do exist on regulated equity tokens, for sanctions and court orders, and their presence is not a defect. What matters is who holds the key and under what published policy it gets used.

Where proof of reserve actually helps

The strongest available construction ties an oracle feed from the custodian's records to a supply cap enforced in the token contract, so that minting beyond attested reserves reverts rather than merely being reported later. That moves the failure from a disclosure problem to an execution problem, which is a real improvement. It still inherits the custodian's honesty and the oracle's liveness, and if the feed stalls the contract has to choose between refusing to mint and trusting a stale number. That choice should be documented. If it is not documented, assume the convenient branch was taken.

The tradeoff nobody states plainly

Tighter backing costs flexibility and yield. Shares that cannot be lent earn nothing from the securities lending market, which is a meaningful revenue line for a custodian and often the thing that lets an issuer charge nothing at the front end. Segregated per-holder accounts cost more to operate than an omnibus account. Continuous proof requires the custodian to expose a feed and accept the operational risk of publishing it. When a product is free and instantly redeemable and promises institutional custody, at least one of those things is being funded by a risk that sits somewhere in the disclosure. Finding where is the whole exercise.

Where it breaks

Three failure modes recur. The first is a chain of intermediaries where the named custodian holds not shares but an interest in another account at a prime broker, so the segregation guarantee only applies at the top layer. The second is a mismatch between the issuing entity and the operating entity, where the attestation covers a special purpose vehicle with clean reserves while the trading obligations sit with a sibling company that has none. The third is corporate action drift, where the reserve was correct before a split or a spin-off and the token supply was adjusted on a different schedule, briefly making the ratio a lie of arithmetic rather than intent.

What to watch

Watch for the words audit and attestation being used interchangeably in marketing, and check which one the report itself claims. Watch the lag between as-of and publication and whether it is shrinking. Watch whether the issuer names the custodian at all, because a disclosure that says a regulated custodian without naming one cannot be independently checked. Watch whether the token contract's mint authority and freeze authority are published addresses. And watch redemption behaviour during stress, because the reserve disclosure is a description and the redemption queue is the test.