Types of stablecoins: the six backing classes and how they differ

Every stablecoin makes the same promise: one token equals one unit of a reference currency, usually the US dollar. What separates them is how they keep that promise. The backing determines what the token holds, who holds it, what can go wrong, and what data you can verify.

The Global Stablecoin Map sorts every coin into one of six backing classes: cash and money-market, other stablecoins, crypto collateral, hedged crypto, own-token and algorithmic, and credit and other real-world assets. Each coin gets exactly one class, set by what the issuer's documents say it is designed to be backed by. Where the reserves have drifted away from that design, the largest reserve share decides. Whether a coin pays its holder is a separate question, covered below.

If you want the fundamentals first, start with what a stablecoin is. This article assumes that ground and focuses on the differences between classes.

These coins hold reserves in bank deposits, cash, short-term government bills, reverse repo and money-market fund shares. An issuer accepts dollars, mints an equal number of tokens, and redeems tokens back for dollars. The peg holds because redemption at par is available to authorized participants, and arbitrage keeps the secondary market price close to that redemption value.

This is the dominant class by supply. USDT, issued by Tether, is the largest stablecoin in circulation. USDC, issued by Circle, publishes monthly reserve attestations and holds most of its reserves in a fund of short-dated Treasuries. PYUSD, issued by Paxos for PayPal, follows the same structure under New York trust company regulation. EURC applies the class to the euro.

The trust assumption is off-chain. Holders rely on the issuer's banking relationships, the quality of the reserve assets, and the attestation or audit regime that reports on them. The reserves themselves are not visible onchain. What is visible is supply, chain distribution, mint and burn activity, and secondary market price. USDC's brief depeg in March 2023, when Silicon Valley Bank held a portion of Circle's cash reserves, showed where this class's exposure sits: not in the token contract, but in the banking layer behind it.

Some coins hold their reserve in other stablecoins, mostly USDC and USDT. The map counts the reserve as the stablecoins it holds, with no look-through to what stands behind them. A coin of this kind is only as sound as the coins it holds, and it adds a second layer of issuer risk on top of the first.

The class includes wrappers and savings products that sit on top of an existing coin. It also includes two well-known names from crypto-collateralized money. DAI and its successor USDS, issued under the Sky brand after Maker's 2024 rebrand, were designed as crypto-collateralized coins. Their reserves drifted from that design, so the largest reserve share decides, and they sit here: Sky's collateral dashboard on 1 October 2026 showed stablecoins held through its peg stability module as the largest bucket of the debt backing them. The class for these coins can change when the mix of reserves moves.

Crypto-collateralized stablecoins replace the bank account with a smart contract. Users lock volatile crypto, typically ETH, staked ETH, or other liquid assets, and mint stablecoins against it at a collateral ratio above 100 percent. If the collateral value falls too far, the position is liquidated and the stablecoin is bought back and burned. Overcollateralization absorbs the volatility of the backing assets.

DAI established the model. Launched by MakerDAO in 2017, it survived the March 2020 crash, though not without a liquidation failure that left the system undercollateralized for a period and forced a debt auction. Liquity's LUSD is the purist version: ETH collateral only, immutable contracts, no governance.

The trust assumption here is onchain and inspectable. Anyone can read the collateral balances, the ratios, and the liquidation parameters directly from the contracts. The trade-off is capital efficiency. Locking $150 or more of collateral to mint $100 of stablecoin is expensive, which is one reason this class holds a fraction of the supply of cash-backed coins. DAI shows a second trade-off: its backing came to include large amounts of USDC and tokenized Treasuries, so the "trustless" design inherited the trust assumptions of the cash-backed class. That drift is why the map now files it under other stablecoins.

Hedged crypto coins hold volatile crypto and cancel its price risk with derivative short positions, so the position as a whole stays worth a fixed number of dollars. The map counts the spot asset and its hedge as one position.

USDf, issued by Falcon Finance, is an overcollateralized synthetic dollar with delta-neutral hedging: the crypto it holds is offset by derivative shorts. Its safety depends on perpetual futures markets staying in positive funding, and on the exchanges and custodians that hold the collateral and the hedge. USDe, issued by Ethena, uses the same design, but in Ethena's backing dashboard on 1 October 2026 liquid stablecoins and DeFi lending outweighed the 19 percent crypto basis, so the map files USDe under other stablecoins for now. The class follows the reserves where they have drifted from the design.

This class covers coins with no collateral, or collateral that consists mainly of the issuing protocol's own token. The classic version pairs the stablecoin with a volatile sister token and lets arbitrageurs mint one by burning the other. When the stablecoin trades below $1, arbitrageurs burn it for $1 of the sister token, contracting supply until the price recovers.

The design has a defining historical event. TerraUSD (UST) grew to roughly $18 billion in supply, paired with the LUNA token. In May 2022, redemptions overwhelmed the mechanism, LUNA's price collapsed under the minting pressure, and UST fell from $1 to under $0.10 within a week. Roughly $40 billion in combined value across the two tokens was erased. Earlier attempts such as Basis Cash and Iron Finance's IRON failed on smaller scales with the same shape: the mechanism works while demand grows and unwinds reflexively when it reverses. Frax is the instructive survivor. It launched as a fractional-algorithmic design, and after May 2022 governance moved it to full collateralization.

The class is small today, and the map's definition is wider than the Terra design. It looks at what sits in the reserve, not at whether the coin runs a mint-and-burn loop. Mento's GBPm, CHFm and JPYm are tagged own-token in the map's data, so they sit here. So does USD0, issued by Usual: its reserve is mostly shares of its own lending vault, plus tokenized Treasury and repo funds. USD0 does not run a Terra-style loop. The tag means that much of its backing is the protocol's own paper. No coin on the map today rebuilds the Terra design at scale. The class stays because the design keeps being attempted, and because knowing the failure mode is part of reading the market.

This class holds loans, private credit, corporate paper, government debt longer than a year, trade finance or real estate. Unlike cash and money-market reserves, these assets cannot be sold at par on short notice, so redemption depends on the issuer's liquidity and on the credit quality of the borrowers.

The class is small on the map. Tokenized Treasury funds do not belong here: they hold short-term government paper, which the map files under cash and money-market, and the map treats the fund shares themselves as investment products rather than stablecoins. The boundary with cash-backed coins is the maturity and liquidity of the assets, and it is genuinely blurry for coins that hold a mix.

The trust assumption is layered. A token onchain represents a claim on an off-chain issuer, which holds instruments at a custodian or lends them out to borrowers. Each layer adds a party. Tokenization changes the auditability of the top layer, not the existence of the layers underneath.

Whether a stablecoin pays its holder is not a backing class. The map records it as its own field, holder return, with three values: none, price accrues, and balance rebases. Most coins pay nothing. A coin whose share price rises against the underlying (a savings wrapper such as sDAI) or whose balance grows without action is flagged as such, whatever backs it.

The yield source defines the exposure. Funding-rate yield depends on perpetual futures markets staying in positive funding. Treasury yield depends on rates staying above zero and on the custody chain behind the tokenized instrument. A savings rate set by governance depends on the protocol's revenue. The advertised number moves constantly, so compare current rates across coins in the stablecoin yields tool rather than relying on any figure quoted in an article.

The table compresses a real gradient, and the class a coin holds can change when its reserves do. The question to ask of any coin is not which box it sits in but what specifically backs it, who controls that backing, and what onchain data exists to check the claim.

Classes describe backing. Supply, chain distribution, peg history, and yield are measurable, and they change weekly. The stablecoin explorer tracks every major coin across these dimensions, with per-coin profiles covering reserves, issuers, and contract addresses, and an issuer view that groups coins by the entity behind them. For how the minting, redemption, and peg mechanics work underneath all six classes, read how stablecoins work.