Algorithmic stablecoins: why the model failed

What is an algorithmic stablecoin and how does it aim to hold 1 dollar?

An algorithmic stablecoin holds no reserves, or only partial reserves, and relies on supply and demand rules coded in smart contracts. When the price slips below 1 dollar, the code creates an incentive to reduce supply. When the price rises above 1 dollar, the code creates an incentive to add supply. The design promises stability from rules instead of from assets.

Reserve-backed tokens answer the question of what stands behind each token with bonds, deposits, or collateral. An algorithmic token answers with a second token and a set of incentives. The ECB suggests treating these designs as unbacked crypto-assets, according to Wikipedia. That label explains why the model stands apart from the rest of the stablecoin family.

How did the mint-and-burn design of the Terra algorithmic stablecoin work?

According to Wikipedia, UST was not backed by US dollars. It used a burn and mint model in which UST was meant to stay stable while the LUNA token absorbed volatility. The model let holders convert UST into LUNA through a mint-and-burn system. When UST traded below 1 dollar, converting it into LUNA was meant to pull the price back up by removing UST from circulation.

The cost sat in LUNA. Each conversion of UST into LUNA increased LUNA supply and pushed its price down, according to Wikipedia. In calm markets the loop stayed invisible, and the design looked efficient. The same loop gave the design its weakness, because the value that absorbed volatility came from a token whose price depended on trust in the project.

Why does the algorithmic model depend on confidence?

The loop is reflexive: the value of the backing token depends on faith in the stablecoin, and faith in the stablecoin depends on the value of the backing token. When both rise, the system looks strong. When one falls, it pulls the other down. A lower UST price meant more conversions into LUNA, more LUNA supply, a lower LUNA price, and less trust in the whole loop.

A reserve-backed token faces a run when holders doubt reserves, yet reserves exist to meet redemptions. An algorithmic token has no equivalent floor, so a run meets nothing that stops it. The only buffer is the market price of the sister token, and that price falls when the run starts. Speed explains the collapse: the loop runs in minutes, with no human step to slow it.

What role did a high yield play in the UST model?

Demand for a stablecoin has to come from somewhere. According to Wikipedia, Anchor depositors who held UST received a 19.45% yield, paid from Terra’s reserves. A yield that high attracted holders who wanted the return, and their deposits supported demand for UST. When demand rests on a subsidy, the subsidy becomes part of the peg.

Wikipedia also records mass withdrawals from Anchor in the days before the collapse. When yield-seekers leave, demand for the token falls just when the loop needs it most. Every stablecoin yield deserves a plain question: who pays it, and what happens to demand when the payment stops? This question applies to every product that pays a return for holding a stable token.

How fast did the algorithmic model fail in May 2022?

UST began to lose its peg on 9 May 2022. Within the following week it fell to 10 cents, or $0.10, and LUNA dropped to virtually zero from an all-time high of $119.51, according to Wikipedia. On 13 May Terraform Labs temporarily halted the Terra blockchain. Almost $45 billion in market capitalization disappeared within one week, a pace that left little time for any rescue.

A bitcoin reserve of 80,394 coins, worth about $2.4 billion on 7 May 2022, stood behind the project, and analysts said on 16 May that its fate remained unclear. On 25 May a proposal was approved to reissue LUNA and abandon UST, and the original chain became Terra Classic. The reserve illustrates the idea of a backstop, and the timeline shows that a backstop must be large and ready before the run begins.

Which other algorithmic designs failed and what do regulators say?

UST is the largest case, and not the only one. Wikipedia lists the Basis project, which shut down and returned funding to investors in December 2018, and a Federal Reserve FEDS Note dated 2 June 2022 that examines runs on algorithmic stablecoins. These cases show a recurring design question: stability without full reserves.

Regulators respond by separating designs. The ECB suggests treating algorithmic tokens as unbacked crypto-assets, and the European MiCA rulebook applies from 30 June 2024 to asset-referenced tokens and e-money tokens. That separation matters for holders: a token that publishes reserves and promises redemption at par sits in a different category from a token that promises stability through incentives.

Which tests reveal the weakness of an algorithmic design early?

Five tests help. What backs the token if confidence falls? Does the sister token have value outside the system? Who pays the yield, and for how long? How fast can the loop run, and can anything slow it? What happened to similar designs in past stress? Answers written with dates give a reading you can repeat each quarter.

A weak answer to two or more questions marks a design that depends on confidence alone. This page is general information, not investment advice. If your treasury in Singapore, ASEAN, or Hong Kong holds a token that matches this profile, the EQOS Asia method page explains the structured scoring, and a 30-minute pre-diagnostic applies it to your case.

Frequently asked questions

About the author

Stéphane Jambu

President, EQOS Forecast Asia

Stéphane Jambu leads EQOS Forecast Asia and works as a consultant in e-commerce, search visibility, and AI. LinkedIn profile