Stablecoin Risks: Six Checks With Dated, Sourced Values

How to use this page

This page does not rank stablecoins by safety. It shows what to check. Six things: the attestation program, the reserve composition, the redemption terms, the licenses, the contracts, and the record of what has happened. Each carries dated sourced values. You apply your own weighting. The checks below are numbered for structure. The numbering is not a ranking. No check is called the one that matters.

Each check ends with a range. The range is what the research found across covered tokens. The range is the information. A class where every token answers alike is a class where the check tells you nothing.

Check one: the attestation program

Does the issuer publish recurring assurance documents? At what tier, from which firm, on what cadence, dated when? The tier names come from the reports’ own language. The range runs wide. Quarterly reasonable assurance under ISAE 3000 Revised, at Tether ^[1]. Monthly examinations under AICPA standards, at Circle ^[2]. Monthly custodian confirmations, at Ethena, which are not audits ^[9]. And one token with no attestation at all. Its verification model is on-chain reads.

Check the tier and its scope. A PDF alone proves nothing. An attestation speaks for its report date. It speaks for its stated assertion. It never speaks for the company as a whole. The attestations page teaches the reading.

Check two: the reserve composition

What backs the token, and how finely is it disclosed? The reports publish composition line by line. The Tether Q2 2026 report holds US Treasury bills at $114.96 billion. It also holds repo, precious metals, Bitcoin, and loan lines ^[3]. The Circle report publishes its own composition lines ^[2]. One protocol publishes collateral through an analytics dashboard. Not an attestation. Its own attribution rules apply.

One phrasing rule rides the word backed. Reserve assets at or above tokens outstanding is the reports’ claim. It never means one hundred percent bank cash. Composition is where the answer lives.

Check three: the redemption terms

The issuer-direct path turns one token back into one dollar. No market sale needed. The terms are published. They differ by orders of magnitude. Tether’s channel works at a 100,000 dollar minimum. Its fees are stated on its schedule ^[4]. Circle’s path runs through a Circle Mint account on published tiers. No daily fee on the Basic tier, up to 40 million dollars. Basis-point steps above ^[5]. A redemption path you have never read is an exit price you are guessing at.

Check four: the licenses

Which registers show which licenses? Register reads dated 11 September 2026, in prose here, show the spread. OCC national trust charters. EU electronic-money authorizations. Japanese registrations. A FinCEN money-services registration at another issuer. One issuer’s public directory shows no entry at all. Its status is issuer-claimed. The honest values are three. Register-verified. Issuer-claimed, with a date. Unresolved. A page flattening those into one word is hiding work.

Check five: the chains and contracts

Where does the token exist natively? Where is it a bridged copy? Circle’s developer docs carry a 37-row mainnet address table, read 11 September 2026 ^[6]. A legacy bridge copy can outlive the issuer’s support for it. The USDC-on-Tron record shows exactly that. A bridged token is not issued or backed by the issuer. Explorers render both kinds alike. The check is short. The issuer’s table. The issuer’s address. Nothing else.

Check six: the operational history

What has happened, how deep, and how did it recover? The record holds four failure modes. No hierarchy is implied. One: terminal collapse, no recovery. The algorithmic case. Two: a bank-access break. The Federal Reserve’s note records the run. Silicon Valley Bank failed after more than 40 billion dollars of withdrawals in a single day. Circle announced 3.3 billion dollars of USDC reserves stuck there. About 8 percent of total reserves. The price troughed at 86 cents, then recovered ^[7]. Three: an allegation-driven depeg, intraday, with redemptions continuing. Four: hours of venue dislocation, primary intact. Four modes. Four lessons. Feeds and pairs named in every case file.

The yield layer

Nothing here is advice. No rate is compared with another. Every figure carries the triple. The value. The fact that it is variable and protocol-set. The as-of date. The Sky interface stated a 3.60 percent savings rate on 11 September 2026 ^[8]. A fixed-term rate read 4.92 percent, dated 26 November. Vault rates read up to 4.88 percent ^[8]. The staking token stUSDS stated 5.17 percent the same day, interface-stated only ^[8]. Ethena’s staked token publishes no static number. The APY is calculated weekly, the docs state, annualized with weekly compounding, positive or flat by design ^[10].

One question is unresolved. Is a given yield instrument a security? This site labels the question. It does not settle it. Regulator positions vary. Several are undeclared. Unresolved is a publishable state. A page implying closure is asserting something no cited document says.

Three misconceptions the checks correct

Attestation equals audit. No. Check one’s tier ladder is the fix. An attestation examines a stated claim. No audit document exists in this corpus. Fully backed means cash. No. Check two’s lines are the fix. Depeg means collapse always. No. Check six shows four modes. Three of the four recovered.

Where this page sits

The stablecoin intro maps the classes whose risks these are, and the cryptocurrency hub is this page’s home root. The USDT and USDC pages carry the two majors’ values for every check, the USDe page carries the custodian model, and is USDT safe applies these six checks to one token in full. The stablecoins hub frames the pillar.