How to Assess a Stablecoin: Reserves, Redemption, Regulation

A stablecoin is a claim on an issuer, not a currency. Assessing one therefore means assessing paperwork rather than technology: what backs the token, who checks that backing and how often, who is legally allowed to redeem it, which authority supervises the issuer, and where the reserves physically sit. Five questions, all answerable from public documents, and the answers differ far more than the shared dollar sign suggests.

What follows is the same rubric we use to score the tokens in our stablecoin ranking, written so you can apply it yourself to something we have not covered.

Question one: what actually backs it?

Three structures dominate, and they fail in completely different ways.

• Fiat-reserve tokens hold cash and short-dated government paper against every unit issued. The risk is the issuer's solvency and the quality of the reserve, and it is largely a credit and custody question.
• Crypto-collateralised tokens are issued against on-chain collateral, usually over-collateralised. The risk is the collateral's price and the liquidation machinery that maintains the ratio under stress.
• Synthetic dollars maintain the peg through a hedged position rather than a reserve. The risk is that the hedge stops working — funding turns persistently against the position, or the venue holding it fails.

None of these is inherently worse than the others. What is worse is presenting one as another. A synthetic dollar described in the language of a fiat-backed one is the failure mode to watch for, and the specificity of the issuer's own language is usually the tell: an issuer that names its mechanism plainly is telling you something, and an issuer whose homepage says only "fully backed" is telling you something too.

Where the backing is on-chain, the excess collateral is generally published and checkable in real time. Where it is off-chain, you are relying on the second question.

Question two: who checks the reserves, how often, and how deeply?

Reserve disclosure varies along three axes, and each one is a real difference in what you know.

Cadence. Monthly is the current floor among the more disclosed issuers; quarterly means that for most of the year you are looking at data up to three months old. Any issuer reporting less often than its comparables is telling you disclosure is not a priority.

Who signs it. An attestation names the firm performing it. An issuer that says "audited" without naming anyone has published a claim, not a document — and an attestation is not an audit in any case: it is a point-in-time confirmation against stated criteria, not an opinion on financial statements.

Depth. This is where the range is widest. At one end, a report gives instrument-level detail — every Treasury holding with its identifier, maturity date and fair value — so the reserve can be reconstructed line by line. At the other, a single figure for "cash and equivalents". Both satisfy the word "attested"; only one lets you form a view.

Here is how that spread looks across the tokens in our ranking, taken from each issuer's own published pages.

As published on each issuer's own pages, read 24 July 2026. Every figure is cited on the review it came from. Disclosure practices change, and an issuer improving its reporting is exactly the kind of change this table is meant to make visible.

Question three: who is actually allowed to redeem?

This is the question that separates a claim on an issuer from a token that merely trades near a dollar, and it is the one retail holders most often get wrong.

Direct redemption — handing tokens to the issuer and receiving dollars — is typically available only to verified institutional counterparties above a minimum size. Everyone else exits through the market, at whatever the market offers at that moment. Both routes usually produce a dollar. They behave completely differently in a crisis, which is the only time it matters: the redemption channel is what anchors the peg, and if you are not a party to it, you are relying on arbitrageurs who are.

So read the redemption page for three specifics: who may redeem, at what minimum, and whether the issuer states redemption is always available. Some issuers are unusually direct here — a token redeemable one-for-one on a named platform at any time, or one-for-one from the issuer with an explicit always-available commitment, is a materially stronger claim than silence. And a protocol-issued token may have no issuer counterparty at all by design, which is not a defect but does mean the peg is maintained entirely by collateral and arbitrage.

How is the peg actually held?

Nothing forces a stablecoin to trade at a dollar. What holds it there is arbitrage against the redemption channel, and the mechanism is worth understanding because it explains why question three matters so much.

When the token trades below par, anyone with direct redemption access can buy it cheaply on the market and redeem it with the issuer for a full dollar, pocketing the difference and removing supply until the gap closes. Above par, the same party mints new tokens at par and sells them into the bid. The peg is therefore only as strong as that channel is wide: if redemption is slow, gated, minimum-sized or suspended, the arbitrage stops working exactly when it is needed, and price is left to the order book alone.

This is why a token with narrow institutional-only redemption can still hold its peg perfectly in calm conditions — the arbitrageurs are present and the channel is open — and why the same token behaves unpredictably under stress, when the channel throttles and everyone else is trading against a book with no anchor beneath it. For a crypto-collateralised token the equivalent question is whether the liquidation machinery can clear collateral fast enough in a falling market; for a synthetic dollar it is whether the hedge can be maintained and unwound at the size required.

What is the reserve actually made of?

"Fully backed" is a statement about quantity. Composition is a statement about quality, and it is where the risk in a fiat-reserve token actually lives. Reserve assets are not interchangeable:

• Bank deposits are instantly available and carry the credit risk of the bank holding them. Concentration in a small number of banking partners is a genuine single point of failure, and it is rarely disclosed at the level that would let you assess it.
• Short-dated government bills are the standard core holding: liquid, and short enough that duration risk is minimal. The maturity profile matters — bills maturing in weeks can be sold at par in a hurry; longer paper cannot always be.
• Overnight reverse repurchase agreements are collateralised lending, generally against government paper, unwinding daily. Liquid, with counterparty exposure that depends on the collateral.
• Anything else — secured loans, corporate paper, other assets — is where the questions start. Not because such assets are unsound, but because they cannot be liquidated at par on demand, which is the entire job of a stablecoin reserve.

An instrument-level disclosure lets you check all of this yourself. A single line reading "cash and cash equivalents" asks you to take the categorisation on trust — and the phrase is broad enough to cover assets you might not consider equivalent to cash at all.

Who can freeze your balance?

Almost every centrally issued stablecoin contract includes administrative functions: the ability to freeze an address, seize a balance, or pause transfers. They exist because issuers operating under supervision are required to be able to comply with lawful orders, and they are not a scandal — but they are a property of the asset that holders should know they have accepted.

The evergreen questions are who holds that key, under what published policy it is used, and whether the issuer discloses that the capability exists at all. Protocol-issued and synthetic tokens generally have no such function, which removes the freeze risk and replaces it with the absence of anyone to appeal to. Neither arrangement is strictly better; they are different trade-offs, and the only bad version is the one you did not know about.

Question four: who supervises the issuer?

Supervision is not a guarantee of anything, and it is a poor proxy for how well-run an issuer is. What it does provide is a party other than the issuer with the power to inspect, and a regime the issuer can be held to.

The practical distinction is between an issuer that names a specific authority and a licence — a state banking regulator, a national trust charter, a payments licence under a named regime — and one that names no prudential supervisor at all. Both exist at large scale. Reasonable people reach different conclusions about how much this matters; what is not reasonable is not knowing which you are holding.

Note also that supervision travels by entity, not by ticker. The same brand may issue through several entities in different jurisdictions under different regimes, and the token you hold was issued by one of them. The issuing entity is named in the issuer's own documentation, and it is worth reading which one applies to you.

Question five: where do the reserves sit, and do they survive the issuer?

Reserves held in an account the issuer controls, in the issuer's own name, are part of its estate. Reserves held by a named third-party custodian and legally segregated are structured to survive its insolvency. The difference only ever matters once, and by then it cannot be changed.

Three things to look for: whether the custodian is named, whether the issuer explicitly states the assets are segregated from its own balance sheet, and whether the reserve sits in a registered fund structure rather than plain deposits — a money market fund brings its own regulatory framework and reporting obligations along with it. Silence on custody is common, and we score it as an unknown rather than assuming the worst or the best.

What this rubric deliberately leaves out

Peg history and market depth are not scored. Both are genuinely relevant to a holder, and both would have to be measured rather than read — from price feeds of varying quality, over windows we would choose ourselves. A rubric is only worth having if every criterion is evidenced the same way, so we score what issuers document and leave the rest to the reader rather than mixing verified paperwork with our own estimates. Where an issuer publishes its own peg metric, the review notes it as the issuer's figure.

Yield is also outside this rubric. Where a token pays a return, the source of that return — interest on reserves, lending, or a hedged position — is a separate risk with its own failure mode, and the rate itself moves too fast for an evergreen assessment.

What we cannot verify

We read what issuers publish; we do not audit reserves, and no reader-facing analysis can. An attestation confirms a position on a single day against stated criteria — it does not follow the assets through the rest of the month, and it does not opine on the issuer's other obligations. Nothing above tells you a token will hold its peg. It tells you what its issuer has committed to in writing, who checks that commitment, and what you could enforce if it failed — which is a different question, and the only one the documents can answer.

Sources

• Reserve, attestation, redemption and custody statements are taken from each issuer's own published pages and cited individually on the reviews linked in the table above, read 24 July 2026
• Our scoring rubric for this category — Ratings Methodology