Automated market makers
An AMM replaces an order book with a pool of two (or more) tokens and a pricing formula. The most common is the constant-product formula x * y = k, where price shifts automatically as the ratio of reserves changes with each trade. Liquidity providers deposit both tokens and earn a share of trading fees, taking on impermanent loss if the pair's price ratio diverges from when they deposited.
- •Bigger trades relative to pool size cause more slippage.
- •LP tokens represent a claim on a share of the pool's reserves and fees.
- •Impermanent loss shrinks if the price ratio returns to the deposit-time level.
Lending and borrowing markets
Protocols like Aave or Compound let users supply assets to earn interest and borrow against collateral, usually overcollateralized: you must deposit more value than you borrow. Interest rates float based on utilization — the fraction of supplied assets currently borrowed — rising as a pool gets tighter to pull in more supply.
- •Collateral factor caps how much you can borrow against a given deposit.
- •Health factor tracks how close a position is to liquidation.
- •Liquidators repay unhealthy debt and seize discounted collateral.
Liquidations
If collateral value falls (or debt value rises) enough that a position's health factor drops below one, anyone can trigger a liquidation: repaying part of the debt in exchange for a bonus slice of the borrower's collateral. This keeps the protocol solvent without needing a central credit desk, at the cost of borrowers needing to actively manage their margin.
Stablecoin designs
Fiat-collateralized stablecoins (USDC, USDT) hold off-chain reserves and rely on a centralized issuer to mint/redeem 1:1. Crypto-collateralized stablecoins (DAI) are backed by overcollateralized on-chain vaults and stay pegged partly through liquidations and stability fees. Algorithmic designs try to hold a peg through supply-adjustment incentives alone; several high-profile ones have de-pegged catastrophically because that incentive loop can break under stress.
- •Fiat-backed: simplest, but introduces custodial and regulatory trust.
- •Crypto-backed: trust-minimized but capital-inefficient (overcollateralization).
- •Algorithmic: capital-efficient but historically the most fragile under panic.