Foundational · v0
Compound
Algorithmic money market — interest rates set purely by pool utilization.
TradFi →Money market fund, rate set by utilization
Prerequisites
01 · Concept — what problem does it solve?
A money market with no rate desk: interest rates are a pure function of . Lenders deposit into a shared pool and receive cTokens; borrowers draw from the pool against collateral. When the pool runs dry, rates spike until lenders arrive or borrowers repay — supply and demand cleared by formula, block by block.
02 · Mechanics
- cTokens: interest-bearing receipts. The exchange rate vs. the underlying only ever increases as interest accrues.
- Kinked rate model: rates rise gently up to a target utilization ("", e.g. 80%), then steeply — protecting exit liquidity.
- Collateral factors: each asset has a max borrow ; cross-asset borrowing against a portfolio of collateral.
- COMP governance: token-holder voting on listings, factors, rate models — the template every protocol copied.
03 · Formulas
// utilization
U = borrows / (cash + borrows − reserves)
// kinked borrow rate
r_b = base + U·m₁ U ≤ kink
r_b = base + kink·m₁ + (U−kink)·m₂ U > kink
// supply rate & cToken exchange rate
r_s = r_b · U · (1 − reserveFactor)
rate = (cash + borrows − reserves) / cTokenSupply
U 80%
Rate = 10.00%
Below kink — gentle slope (base + U × m₁)
04 · Edge cases & risks
- 100% utilization — lenders cannot withdraw and liquidators cannot seize; the steep slope above the kink exists to make this state self-correcting.
- manipulation — Nov 2020: DAI printed 30% high on Coinbase Pro feed → ~$89M of positions wrongly liquidatable.
- Token distribution farming — COMP emissions briefly made borrowing net-profitable, inflating utilization with circular positions.
- Governance latency — a buggy Proposal 62 (2021) mis-distributed ~$80M COMP; the timelock meant no fast fix existed.