An algorithmic stablecoin is a token that tries to hold a stable price, usually one US dollar, by expanding and contracting its supply through programmed rules and market incentives rather than by holding a full reserve of cash or collateral for every unit. Among stablecoins, it is the design that relies most on market confidence and least on assets that holders can actually redeem.
In crypto & blockchain, the category covers several mechanisms: seigniorage and two-token models, rebasing tokens and fractional-algorithmic hybrids. What unites them is that the peg is defended by arbitrage and reflexive incentives, not by a custodian holding dollars or by overcollateralized vaults.
How an algorithmic stablecoin defends its peg
An algorithmic stablecoin defends its peg by making it profitable for traders to push the price back to target. When the token trades above one dollar, the protocol lets users create new units cheaply and sell them, which adds supply and lowers the price. When it trades below one dollar, the protocol offers something worth more than the market price in exchange for burning units, which removes supply and raises the price.
In the classic two-token design, a second volatile token absorbs the swings. Users can always swap one stablecoin for one dollar worth of the governance or share token, and the reverse. The share token therefore acts as implicit equity in the system: its market value is the only thing standing behind the promise to redeem at par.
The weakness is reflexivity. If the stablecoin slips below the peg and many holders redeem, the protocol mints large amounts of the share token, its price falls, each new unit buys less, and confidence drops further. This feedback loop is known as a death spiral, and it can unfold within days or even hours.
The table below shows how the main mechanisms respond when the market price moves away from the target.
| Mechanism | Response above the peg | Response below the peg |
|---|---|---|
| Two-token mint and burn | Burn share token to mint stablecoin | Burn stablecoin to mint share token |
| Rebasing supply | Every wallet balance grows | Every wallet balance shrinks |
| Seigniorage shares and bonds | New supply paid to shareholders | Discounted bonds sold to remove supply |
| Fractional-algorithmic hybrid | Collateral ratio may be lowered | Collateral ratio raised, buybacks |
Design variants and their parameters
Designers tune a small set of control parameters that determine how aggressively the system reacts. The most important are the collateral ratio, the size of the mint and redemption spread, the speed of supply changes and the source of the price oracle that tells the contract whether the token is above or below target.
| Parameter | Typical setting | What it affects |
|---|---|---|
| Collateral ratio | 0 to 100 percent | How much hard value backs each unit |
| Redemption fee or spread | Fractions of a percent | Profit left for arbitrageurs |
| Rebase interval | Hours to one day | Speed of supply correction |
| Oracle window | Minutes to hours of averaged prices | Resistance to price manipulation |
| Mint or burn caps | Daily limits, if any | Protection against sudden runs |
A rebasing token changes every holder’s balance proportionally, so the price per unit tends toward target while each holder’s share of total supply stays constant. This keeps the unit price near one dollar but does not preserve the value of a holding, which is why rebasing designs rarely function as practical money.
Fractional-algorithmic models hold some collateral and cover the rest with a share token. They are generally less fragile than pure designs, but the uncollateralized portion behaves exactly like a pure algorithmic stablecoin under stress, so a partial reserve only delays a run if confidence collapses.
How it differs from collateralized stablecoins
The defining difference is what stands behind redemption. A fiat-backed stablecoin promises redemption against bank deposits and short-term government debt held by an issuer, and a crypto-collateralized stablecoin locks more crypto value than it issues. An algorithmic stablecoin promises value that it largely creates from its own token economy.
| Type | Source of backing | Main risk |
|---|---|---|
| Fiat-backed stablecoin | Cash and short-term government debt | Issuer, custody and banking risk |
| Crypto-collateralized stablecoin | Overcollateralized crypto in vaults | Collateral crash and liquidation failure |
| Algorithmic stablecoin | Protocol rules and a native token | Loss of confidence and death spiral |
| Fractional hybrid | Partial collateral plus native token | Run on the uncollateralized share |
Because the backing is endogenous, an algorithmic stablecoin can scale quickly and needs no bank, which explains its appeal. That same property means the backing shrinks exactly when it is needed most. A collateralized design can also depeg, but it usually has assets that can be sold to restore redemptions.
Failure modes and lessons from collapses
Several high-profile algorithmic stablecoins have lost their peg permanently, and the collapse of TerraUSD in May 2022 remains the defining case. Its mint-and-burn link to the LUNA token turned large redemptions into hyperinflation of LUNA, and the stablecoin fell far below one dollar within about a week. Earlier experiments such as Basis Cash and Iron Finance failed in similar ways.
- Unsustainable yield: high deposit rates attract capital that leaves the moment yields fall.
- Concentrated liquidity: a few large pools let a big seller move the price sharply.
- Oracle lag: slow or manipulable price feeds let attackers mint or redeem at wrong prices.
- Reserve illusions: emergency reserves in volatile assets lose value during the same crash.
- Governance delay: parameter changes arrive after the run has already started.
These episodes changed regulation. Frameworks such as MiCA in the European Union and the GENIUS Act in the United States favor stablecoins with full reserves and a legal redemption right, and they leave little room for designs that rely only on algorithms. Purely algorithmic tokens may still exist on public blockchains, but they generally cannot be marketed as regulated payment stablecoins.
Evaluating an algorithmic stablecoin
Anyone assessing such a token should start from one question: what can a holder receive if everyone redeems at once. The answer reveals how much of the peg rests on real assets and how much on expectations.
| Signal | What it indicates | Prudent response |
|---|---|---|
| Low or zero collateral ratio | Peg rests on confidence alone | Treat the token as speculative |
| Double-digit fixed yield | Growth funded by subsidies | Ask where the yield comes from |
| Persistent trading below par | Arbitrage is not closing the gap | Reduce exposure early |
| Fast share token inflation | Redemptions are diluting backing | Expect a possible spiral |
Stripped to the essentials
An algorithmic stablecoin is a digital token that tries to stay worth one dollar without keeping one dollar in a vault for every token. Instead, a program adds tokens when the price is too high and removes tokens when it is too low, and it pays traders a small reward for helping.
This works while people believe in the system. The weak point is trust: if many holders want out at the same time, the reward the program offers is paid in a token that is itself losing value, and the price can fall very quickly. It resembles a club that promises refunds paid in club membership points.
Where it is applied
A treasury team at a mid-sized decentralized finance protocol reviews which stablecoins it may hold for operating expenses. One candidate is an algorithmic stablecoin offering a 20 percent deposit yield, backed by a native share token and a reserve worth roughly 15 percent of supply.
The team models a scenario in which a third of holders redeem within 48 hours. Redemptions would mint new share tokens far faster than the market could absorb them, and the reserve would cover only a small part of the outflow. The team concludes that the token behaves like a leveraged bet on its own share token, not like cash.
The protocol keeps operating funds in a fully reserved, regularly attested stablecoin and allows only a small, capped speculative allocation to the algorithmic token. When the token later trades at 0.94 dollars for several days, the loss stays within the planned limit.