Algorithmic Stablecoins: Lessons from Past Failures
Algorithmic stablecoins try to hold a peg with code instead of full reserves. Learn how they work, why TerraUSD and others collapsed and the warning signs.
An algorithmic stablecoin is a cryptocurrency that tries to hold a fixed value — usually one US dollar — mainly through code-driven supply adjustments and trading incentives, rather than by holding a full reserve of cash, bonds or other assets. When demand falls, the system tries to shrink supply; when demand rises, it expands supply.
The idea is appealing: a dollar-like token that doesn't depend on a bank or a company holding reserves. But the history of algorithmic stablecoins is mostly a history of failures, including TerraUSD's collapse in 2022, one of the largest losses in crypto history. Understanding why they break helps you judge any stablecoin's claims.
How Algorithmic Stablecoins Are Supposed to Work
Most stablecoins fall into broad categories (see our overview of stablecoins):
| Type | What backs it | Example of the approach |
|---|---|---|
| Fiat-backed | Cash and short-term government debt held by an issuer | Major dollar stablecoins issued by companies |
| Crypto-collateralized | More crypto locked than stablecoins issued | DAI |
| Algorithmic | Mainly code and incentives; little or no external collateral | TerraUSD (UST) before its collapse |
| Hybrid / fractional | Partly collateral, partly algorithmic | IRON (Iron Finance) |
Common algorithmic mechanisms include:
- 1Two-token mint and burn. Users can always swap one stablecoin for one dollar's worth of a companion token, and vice versa. If the stablecoin trades at $0.98, traders buy it and redeem for $1 of the companion token, reducing supply and lifting the price.
- 2Rebasing. The protocol changes every holder's balance up or down to move the price toward a target.
- 3Bonds or coupons. When below peg, users buy discounted "bonds" that pay out later when the peg recovers.
All of these depend on one assumption: that someone will always be willing to step in at a discount because they believe the system will recover.
Why They Fail: The Death Spiral
The weakness shows up when confidence drops:
- 1The stablecoin slips below $1 because many holders want out at once.
- 2Arbitrage redemptions mint large amounts of the companion token.
- 3New supply floods the market, so the companion token's price falls.
- 4Each redemption now requires even more companion tokens, accelerating the fall.
- 5Holders realize the "backing" is only as valuable as a collapsing token, and rush to exit.
In other words, the system uses its own token as collateral. When that token's value depends on confidence in the stablecoin, both can fall together.
Lessons from Past Failures
TerraUSD (UST) and LUNA, 2022
UST was the largest algorithmic stablecoin. It used the two-token model with LUNA, and a lending protocol called Anchor offered a high, steady yield on UST deposits, attracting enormous demand. In May 2022, large withdrawals and sales pushed UST below its peg. Redemptions minted vast amounts of LUNA, its price collapsed, and both tokens lost nearly all their value within days. Tens of billions of dollars of market value disappeared, and the failure spread to lenders and funds across the industry.
The original chain was later renamed Terra Classic, and LUNA became LUNC — covered in our Terra Classic explainer.
Lesson: a high, unexplained yield can create demand that isn't sustainable, and reflexive two-token designs can unwind with extreme speed.
Iron Finance (IRON), 2021
IRON was partly backed by USDC and partly by the project's own token, TITAN. In June 2021, heavy selling of TITAN triggered a bank-run-style spiral: redemptions minted more TITAN, its price fell toward zero, and IRON lost its peg.
Lesson: even partial collateral doesn't save a design if the remaining portion is the project's own volatile token.
Earlier experiments
Projects such as NuBits and a series of "seigniorage" and rebase-based stablecoins in 2020 and 2021 also failed to hold their pegs once demand faded. Several projects that started with partly algorithmic designs later moved toward full collateral.
How to Evaluate a Stablecoin's Design
Ask these questions before holding or earning yield on any stablecoin:
- What backs it? External assets (cash, treasuries, ETH), or the project's own token?
- Is it over-collateralized? If collateral falls, what happens — liquidations, or reliance on new buyers?
- Who can redeem, and at what price? Can you redeem directly, or only sell on the market?
- Where does the yield come from? If you can't explain it, assume risk is being hidden.
- How did it behave under stress? Look at its history on our historical data pages.
- How concentrated is the supply? A few large holders can trigger a run.
Our guide on evaluating a stablecoin's backing and redemption gives a fuller checklist.
Newer Designs Are Not Risk-Free
Some modern "synthetic dollars", such as Ethena's USDe, are not algorithmic in the Terra sense — they hold crypto collateral and use hedging positions. They carry different risks: exchange and custody exposure, funding-rate changes and liquidity during market stress. Crypto-collateralized stablecoins like DAI depend on collateral values, oracles and liquidations working as intended. "Not algorithmic" doesn't mean "can't depeg". See stablecoin depegs.
Warning Signs Checklist
- Backing consists mostly of the project's own governance or "share" token.
- Unusually high, fixed yields on the stablecoin.
- Marketing that says "always worth $1" without explaining redemption.
- No independent reporting on collateral.
- Large share of supply in one protocol or a few wallets.
Bottom line
Algorithmic stablecoins promise stability without reserves, but in practice their "backing" is confidence — and confidence can vanish in hours. TerraUSD and Iron Finance showed how quickly a peg can collapse when redemptions inflate a companion token. Judge any stablecoin by what truly backs it, how redemption works and where its yield comes from, and don't treat any token as risk-free just because it's labeled "stable".
Frequently asked questions
What is an algorithmic stablecoin?
An algorithmic stablecoin tries to keep a stable value, usually one US dollar, mainly through code-driven supply changes and market incentives rather than holding full reserves of cash or other assets. Typically the system mints or burns tokens, or a companion token, to push the price back toward the peg.
Why did TerraUSD (UST) collapse?
UST relied on traders being able to swap one UST for one dollar's worth of newly minted LUNA. When large withdrawals and sales pushed UST below its peg in May 2022, redemptions created huge amounts of LUNA, crashing its price. That destroyed confidence in the mechanism, and both tokens lost almost all their value within days.
Are all algorithmic stablecoins doomed to fail?
Not necessarily, but designs that rely mainly on the value of their own companion token have repeatedly failed under stress. Many projects now use full or over-collateralization with external assets. Any stablecoin can lose its peg, so check what backs it and how redemptions work.
Is DAI an algorithmic stablecoin?
DAI is usually described as a decentralized, crypto-collateralized stablecoin rather than a purely algorithmic one. It is issued against collateral that must exceed the amount of DAI created, and positions are liquidated if collateral falls too far. That backing distinguishes it from designs like TerraUSD.