Algorithmic Stablecoins and Seigniorage Designs

A dollar-backed stablecoin is simple to reason about: hold a dollar in a bank account for every token issued, and the peg holds as long as the dollars are really there. An algorithmic stablecoin tries something more ambitious — hold no reserves at all, and instead keep the price at $1 purely through a built-in incentive to arbitrage it back whenever it drifts, using a second, unbacked token as the shock absorber.

The seigniorage mechanism

Picture a two-token system: a stablecoin (call it USD-X) meant to hold $1, and a companion token that absorbs volatility (call it VOL). The protocol lets anyone burn $1 worth of VOL to mint 1 USD-X, or burn 1 USD-X to mint $1 worth of VOL, at the current market price of VOL.

If USD-X trades above $1, arbitrageurs are incentivized to burn VOL and mint new USD-X — since minting costs less than $1 of VOL but produces a token worth more than $1 — and sell the new USD-X, pushing its price back down. If USD-X trades below $1, the reverse: burn cheap USD-X, mint VOL worth a full dollar's face value, and sell it, which should shrink USD-X supply and push its price back up.

Worked example: the loop working as designed

USD-X trades at $0.98. An arbitrageur burns 100 USD-X (worth $98 at market) and receives $100 worth of VOL at VOL's current price, say $10, meaning they receive 10 VOL tokens. They immediately sell the VOL for close to $100 in cash on the open market.

1. Cost to the arbitrageur. $98 to acquire and burn the 100 USD-X.
2. Proceeds. ~$100 from selling the newly minted VOL.
3. Profit. $100 - 98 = 2$, or $2, and the act of burning 100 USD-X shrinks its circulating supply, which — if demand for USD-X is unchanged — should push its price back toward $1.

Worked example: the loop running backward

Now suppose confidence in VOL itself is shaken — perhaps USD-X has depegged sharply and everyone knows more VOL is about to be minted to defend it. USD-X trades at $0.80. An arbitrageur burns 100 USD-X (worth $80) and mints $100 worth of VOL at the stated formula — but when they try to sell that VOL, the market, flooded with sellers doing the exact same trade, only pays $40 for it.

1. Cost. $80 to burn the USD-X.
2. Actual proceeds. $40 — far below the $100 face value the formula promised, because VOL's price is collapsing under the sudden new supply.
3. Result. A $40 loss instead of a profit, and every unit of VOL minted this way adds more sell pressure, which pushes VOL's price down further, which requires minting even more VOL to absorb the next unit of USD-X — a hyperinflationary spiral in the companion token that ends near zero, taking the "stable" coin down with it once the promise of full $1 redemption is no longer credible.

What this means in practice

Algorithmic designs looked capital-efficient — no idle collateral sitting around — right up until a large-scale depeg event, most visibly the 2022 collapse of a major algorithmic stablecoin and its companion token, which went from tens of billions in combined market value to near zero within days. The mechanism that defends the peg in normal times is the same mechanism that destroys both tokens in a crisis, because the "collateral" backing the stablecoin is just market confidence in a second token, not an independent asset.