Algorithmic stablecoins hold a $1 peg through code and incentives rather than fully backed reserves. When those incentives fail, the peg can break in hours, and holders discover there's no vault of cash behind the token. TerraUSD's 2022 collapse remains the clearest case study, and its lessons are the fastest way to evaluate any new "stable" design you're considering.
How the TerraUSD collapse actually worked
UST held its peg through a mint-and-burn relationship with LUNA, a volatile sister token. To create one UST, you burned a dollar's worth of LUNA. To redeem UST, you minted LUNA. In calm markets, arbitrageurs kept the price near $1.
In a panic, the mechanism became a death spiral. As UST slipped below peg, holders redeemed for LUNA, which pushed LUNA's price down, which meant each UST was worth less LUNA, which pushed UST down further. The design had no external collateral. Confidence was the collateral.
Lesson one: "backed by math" is a marketing phrase
A stablecoin is only as stable as the assets standing behind it. Reserve-backed coins hold cash, Treasuries, or short-term debt. Algorithmic designs hold a promise that arbitrage will restore the peg. That promise works until enough people want out at once.
Questions to ask about any reserve claim
- What exactly backs the token, and who holds it?
- Can I redeem directly, or only sell on a secondary market?
- Is there an independent attestation, and how recent is it?
- What happens to the peg if the backing asset falls in price?
If the answer to the first question is "the token's own ecosystem," you're looking at reflexivity, not collateral.
Lesson two: reflexivity turns a wobble into a spiral
A reflexive system depends on its own price to stay stable. That's fine on the way up and catastrophic on the way down. Every redemption makes the next redemption more attractive.
Suppose you hold $10,000 in a reflexive stablecoin. It slips to $0.97 and you decide to wait for recovery. Within a day it's at $0.80, then $0.60. By the time you try to exit, the secondary market is thin and the discount is brutal. A holder who sold at $0.97 lost $300. A holder who waited lost thousands. The lesson isn't "sell at the first dip" — it's that a reflexive peg gives you no floor to catch you.
Spotting reflexivity before it bites
Look at what has to happen for the peg to hold. If the answer involves "the other token goes up," the system is reflexive. If it involves "someone redeems for cash," it's collateralized. One depends on sentiment, the other on assets.
Lesson three: yield that looks too good is a warning, not a bonus
UST's ecosystem offered eye-catching yields on deposits, which pulled in capital that had no other reason to be there. High yield on a stablecoin usually means the protocol is paying you to hold risk it can't otherwise sell. That's not automatically fraud, but it's a signal to look harder at where the return comes from.
| Design | What holds the peg | Main failure mode |
|---|---|---|
| Fiat-backed | Cash and short-term reserves | Custody or disclosure risk |
| Crypto-backed | Overcollateralized crypto | Liquidation cascades in a crash |
| Algorithmic | Arbitrage and incentives | Loss of confidence, death spiral |
None of these is risk-free. The point is knowing which risk you're actually taking.
Lesson four: exits matter more than entries
During the UST unwind, the bottleneck wasn't buying — it was getting out. Order books thinned, redemption queues formed, and the discount widened. Any stablecoin you use should be judged on how it behaves when everyone wants the door at once.
A practical stress test
Ask yourself: if 20% of holders tried to exit this week, what would happen to the price? For a reserve-backed coin, the answer involves redemption at par. For an algorithmic one, it involves a slide. Run that test before you park savings, not after.
Where AI tooling fits — and where it doesn't
Analytics platforms and AI-assisted dashboards can surface peg deviations, redemption volumes, and reserve disclosures faster than manual checks. That's genuinely useful for monitoring. It doesn't make a weak design safe. Tools like MGBX and similar research interfaces help you compare structures, but the judgment call — is this token backed by assets or by hope? — stays with you. No dashboard replaces reading the reserve section.
Frequently asked questions
Are all algorithmic stablecoins doomed?
No design is automatically doomed, but purely reflexive models have repeatedly struggled under stress. Some newer designs add partial collateral or circuit breakers to slow a spiral. The key question is whether there's a real asset buffer when confidence drops, because confidence alone has proven fragile.
How fast can an algorithmic stablecoin lose its peg?
It can happen within hours once redemptions outpace the mechanism's ability to absorb them. The UST unwind unfolded over days, but the decisive damage occurred quickly as the reflexive loop kicked in. Speed is part of the risk — you may not have time to react once the slide starts.
What should I check before holding any stablecoin?
Start with what backs it and who custodies the reserves. Then look at redemption terms and whether independent attestations exist. Finally, ask how the peg behaves if a large share of holders exit at once. If any of those answers is vague, treat the token as higher risk than its $1 price suggests.
Next time you see a stablecoin promising high yield and a "mathematically guaranteed" peg, open its reserve documentation before you deposit a dollar.