An algorithmic stablecoin tries to hold a fixed value, usually one dollar, using a code rule and a paired token instead of real backing. Software mints new coins or burns them to nudge the price back toward the target. There is little or no reserve to redeem, so the design leans on confidence.
The catch is that the design is circular. The coin's peg leans on a second token, and that token's value leans on the coin. While demand rises, the loop looks steady and even clever, but when demand falls it can run in reverse and spiral fast. That circular dependence is why an algorithmic coin sits at the far, riskiest end of the main types of stablecoins. If the basics are still fuzzy, start with what a stablecoin is.
TerraUSD is the lesson every later design is measured against.
How an algorithmic stablecoin holds its peg
An algorithmic stablecoin holds its peg with a rule written in code, not with cash in a bank. When the price rises above a dollar, the rule adds new coins to push it down. When the price falls below a dollar, the rule shrinks the supply, often by paying people to swap into a partner token.
Think of it as a tiny central bank that runs on autopilot. A real central bank can print money or pull it back to steer a currency, and an algorithmic coin does the same with a smart contract. It grows the supply when the coin is too expensive and shrinks it when the coin is too cheap, holding few or no assets in reserve (source: Federal Reserve, The stable in stablecoins). Analysts often sort stablecoins into two broad camps for this reason: fully reserved coins that hold real assets, and algorithmic coins that lean on a rule (source: J.P. Morgan Global Research on stablecoins). The goal is to keep each coin near a dollar without holding a dollar behind it.
The below-peg case is where the paired token earns its keep. Say the coin slips to $0.90. The rule lets you hand in one coin and receive a dollar's worth of the partner token, or buy a bond now that pays out a full dollar once the peg returns. Either way, you burn the stablecoin, so its supply shrinks and the price is nudged back up. Above a dollar, the rule runs in reverse: it mints fresh coins and hands them to partner-token holders, adding supply until the price eases down.
How a peg actually snaps back, through redemption and the profit motive that traders bring, is a shared topic across every design, and how stablecoins hold their price covers that machinery in full. These coins also run as programs on public blockchains, so how DeFi apps work is useful background for where the rule actually lives. All of it works smoothly while one assumption holds.
Reflexivity: why the design leans on confidence
The design leans on confidence because the coin and its partner token prop each other up. The stablecoin's peg depends on the partner token holding value, and the partner token holds value because people trust the stablecoin. That circle has a name, reflexivity, and it is the design's core weakness. Break the trust, and both fall together.
Here is a plain way to picture it. Imagine you write an IOU for a dollar, then promise that if anyone doubts it, you will cover it with a second IOU that you can also print whenever you like. While people accept the second IOU, the first one looks solid. The moment they stop, you are printing paper to back paper, and neither is worth a dollar. An algorithmic stablecoin is close to that: the "reserve" it points to is a token it can create at will.
Compare that with real backing. A cash-backed coin has a dollar in a bank whether or not anyone trusts it that day. A crypto-backed coin locks up more crypto than the coins it issues, so a buffer absorbs the swings. An algorithmic coin has neither. It has a rule and a partner token, and that partner token is usually as volatile as any other crypto.
The loop cuts both ways, which is what makes it dangerous. While demand grows, minting the partner token spreads value around and the whole thing reinforces itself upward. When demand shrinks, it reinforces itself downward: the rule mints more of the partner token to defend the peg, that extra supply drives the partner price lower, and the lower price means even more must be minted. The families below differ mostly in how hard they lean on this loop.
The main families of algorithmic stablecoins
Algorithmic stablecoins come in three main families. Rebase coins change the balance in every holder's wallet. Seigniorage or two-token coins pair the stablecoin with a separate share token. Fractional or hybrid coins mix a little real collateral with the algorithm. All three steer price by moving supply, leaning on the same confidence loop.
A rebase coin has an elastic supply. If the coin trades above target, the rule raises everyone's balance at once; if it trades below, it lowers everyone's balance. Your share of the network does not change, but the token count in your wallet does, which can feel strange. Ampleforth is the best-known rebase design, and it aims for a steady price per token rather than a hard dollar peg (source: Ampleforth documentation).
A seigniorage or two-token coin keeps your balance fixed and pushes the work onto a second token. Below peg, you can buy a bond cheaply and redeem it for a dollar once the coin recovers, which pulls coins out of circulation. Above peg, new coins are minted and handed to share-token holders. TerraUSD with its LUNA token, and the earlier Basis Cash, worked this way. Fractional or hybrid coins hold part real collateral and part algorithm, like the early version of Frax, though most of these moved toward fuller backing after 2022 because the algorithmic slice brings the loop back (source: Frax Finance, original design).
| Family | How it holds the peg | What it leans on | How it breaks | Example |
|---|---|---|---|---|
| Rebase (single token) | Changes the token balance in every wallet to steer price | Holders accepting a moving balance, plus accurate price feeds | Your balance shrinks in a downturn while the price still drifts | Ampleforth |
| Seigniorage / two-token | Mints or burns a paired share or bond token to move supply | Steady demand for the paired token | The loop reverses, the paired token floods, and both fall | TerraUSD with LUNA |
| Fractional / hybrid | Part real collateral, part supply algorithm | The collateral share plus trust in the algorithmic part | The algorithmic slice reintroduces the loop under stress | Frax (early design) |
One collapse shows this loop in motion better than any diagram.
What TerraUSD showed about the design
TerraUSD, or UST, is the design lesson every algorithmic coin is measured against. It held its dollar peg with a swap into a partner token called LUNA, letting one UST turn into a dollar's worth of LUNA and back. When enough holders rushed for the exit in May 2022, that swap turned the loop against itself.
The mechanism was a simple swap. One UST could always be turned into a dollar's worth of freshly minted LUNA, and the other way around, which was meant to hold UST near a dollar (source: Richmond Fed brief on the Terra collapse).
Demand for UST had been propped up by yield. A lending app called Anchor paid about 19.5 percent to hold it, far above safe rates, so money poured in while chasing how DeFi yield works. A payout that high is hard to keep going, and it carries the risks of chasing yield that come with any rate far above the norm. It left the system resting on people who were there for the payout, so when confidence cracked, the swap rule became the problem instead of the fix.
Reflexive loop, simplified (TerraUSD, May 2022) 1. Holders lose confidence and sell UST -> price slips below $1 2. To defend the peg, the rule lets holders swap 1 UST for $1 of newly minted LUNA 3. That fresh minting floods the market with LUNA -> LUNA's price falls 4. A lower LUNA price means even more LUNA must be minted for each UST -> supply explodes 5. LUNA's supply ballooned about 80 times in two days, from roughly 0.4 billion to 32 billion tokens, as its price crashed from about $31 to about a cent 6. With no reserve to redeem against, the peg had no floor, and both tokens fell to near zero
Those supply figures come from the Richmond Fed's account, which also notes that about $18 billion of UST was in circulation at the peak. That value collapsed within about a week. Because an algorithmic coin holds no reserve to redeem against, there was nothing to recover to, so the peg did not bounce back the way a fully backed coin's can.
That is the opposite of an ordinary wobble. A fully backed coin that slips because its cash is stuck at a bank for a weekend can recover, because the real dollars still exist. UST had no dollars to return to. What to actually do if a coin you hold starts to slip (the signals to watch, the choices to make) is its own topic: stablecoin depeg risk. This guide stays on the design, not the response. The collapse was large enough that rule-makers stepped in.
Why regulators treat pure-algorithmic coins differently
After Terra, regulators began treating pure-algorithmic coins as their own, higher-risk category rather than as ordinary stablecoins. The reason is simple. A coin with no real reserve cannot meet a full-backing standard, and new rules for payment stablecoins increasingly demand assets a holder can actually redeem, which a printed-token peg does not have.
In the United States, the GENIUS Act, signed in 2025, will require a payment stablecoin to be fully backed by high-quality liquid assets such as cash and short-term government debt once its rules take effect (source: GENIUS Act, Public Law 119-27). A pure-algorithmic coin holds none of that, so it will not qualify as a regulated payment stablecoin in that market. One detail matters for timing: the law is signed, but its detailed rules are still being written, so it is not fully in force yet. The logic behind the rule is the one Terra proved. If a coin cannot point to assets it is able to sell, it cannot honor a dollar when the pressure comes.
The contrast is the whole point. The Bank of England describes a stablecoin as something whose issuer holds the same value in real money, with a right for holders to swap back on demand (source: Bank of England stablecoin explainer). A pure-algorithmic coin has neither the reserve nor a real redemption right, and that gap is exactly what regulators singled out. The precise rules differ by country and keep changing, so the jurisdiction-by-jurisdiction detail is its own subject; the takeaway here is about the design, not the law. Rules aside, the practical job for a reader is to spot the design in the first place.
How to recognize and avoid a pure-algorithmic stablecoin
To recognize a pure-algorithmic coin, ask three things: does its peg lean on a second token, is there a real reserve you could redeem, and is an unusually high yield pulling people in. If the answers are a partner token, no reserve, and a rich payout, treat it as high risk and, for most people, avoid it.
A short routine catches almost every case.
- Ask what backs it. If the honest answer is "a second token this same project also issues," there is no outside backing to fall back on when trust drops.
- Look for a real reserve you could redeem. A cash-backed or crypto-backed coin can point to assets it holds; a pure-algorithmic coin can only point to a rule.
- Be wary of a high, advertised yield. If a coin has to pay well above safe rates to get people to hold it, that yield is quietly recruiting the demand the peg depends on, which is the pattern that preceded the biggest failure.
- Check where it trades and who runs it. These coins live inside on-chain apps, and the split between centralized and decentralized exchanges changes who holds your coins and what can go wrong.
- Size any position to the risk. If you still want exposure, keep it small, the same way you would weigh risk and return for anything this volatile.
From BloFin's operational view, the lasting lesson here is a simple check, not a judgment call: a dollar token that leans on a second token to hold its peg carries a design risk that even deep liquidity cannot fix. On BloFin's platform, the stablecoins that carry real day-to-day settlement are the ones backed by assets a holder can redeem, not by a partner token the system prints, and a real-backing product like RWUSD sits on the opposite side of that line from a pure-algorithmic coin. An algorithmic design can be elegant on paper, but for most people it is a thing to recognize and avoid, not a place to park money.
Frequently asked questions
Are any algorithmic stablecoins still around today?
A few still trade, but they are a small slice of the market, and most survivors are hybrids that hold real collateral rather than pure-algorithmic coins. Designs with no reserve have largely fallen out of favor since 2022. Some rebase coins still exist, and some projects keep a partial-collateral model with circuit breakers to slow a fall. The pattern is consistent: the closer a coin gets to full backing, the steadier it tends to be, and money has moved that way.
Is a synthetic dollar like USDe an algorithmic stablecoin?
No, and the mix-up is common. A synthetic dollar such as USDe holds real crypto collateral plus an offsetting short position, so its backing sits outside the token and can be sold if needed. An algorithmic coin has no such reserve, since it leans on a partner token it can print. Both can look code-run from the outside, but one holds outside assets while the other holds a promise. That difference in backing is the whole point, and it is why USDe is grouped with synthetic designs, not algorithmic ones.
What is the difference between a rebase and a seigniorage stablecoin?
Both steer price by changing supply, but in different places. A rebase coin changes the token balance in every wallet at once, so your token count moves up or down while your share of the network stays the same. A seigniorage or two-token coin leaves your balance alone and instead mints or burns a separate share or bond token to move supply. In short, rebase spreads the adjustment across all holders, while seigniorage pushes it onto a paired token that is meant to absorb the swings.
Why did people trust TerraUSD if it had no reserves?
Mostly because of yield and momentum. A lending app paid about 19.5 percent to hold UST, which is far above safe rates, so demand poured in while the peg held. A rising price and a large ecosystem made the risk easy to overlook. That is the trap with a confidence-based design: it can look safest right before it breaks, because the same demand that props up the peg also hides how little sits behind it. When the demand turned, there was nothing underneath.
Can an algorithmic stablecoin ever be made safe?
Not fully, as long as it stays pure-algorithmic. The weakness sits in the design, not in one team's execution. Adding partial collateral, circuit breakers, or supply caps can slow a fall, and some hybrids have lasted longer because of it. But a coin with no real reserve still has nothing to redeem against once confidence goes. The telling detail is that the safer these designs get, the more they start to resemble a backed coin, which is really an admission that pure algorithms are not enough.
Are algorithmic stablecoins banned now?
Not outright in most places, but the rules are tightening. New payment-stablecoin laws require full backing in safe assets, which a pure-algorithmic coin cannot meet, so it cannot qualify as a regulated payment stablecoin in those markets. Owning one is rarely the legal problem; the rules mostly bite on who is allowed to issue and market a stablecoin, so the practical effect is fewer new pure-algorithmic coins reaching everyday users.
Researched and written by the BloFin Academy editorial team with AI-assisted drafting. Updated July 2026. Primary sources: the Federal Reserve, the Richmond Fed, the Bank of England, the GENIUS Act (Public Law 119-27), and protocol documentation from Ampleforth and Frax. All facts independently verified against cited sources current as of July 2026.
This article is educational and general in nature, not financial, investment, tax, or legal advice. Algorithmic stablecoins carry a high risk of losing their peg and, in the worst case, of near-total loss, and past collapses show the value is not guaranteed. Nothing here is a recommendation to buy, sell, or hold any specific asset. Do your own research, and consider a licensed professional before making financial decisions. BloFin does not provide investment advice.