Onchain Atlas

Overcollateralization

Requiring borrowers to lock up more value than they borrow, so the lender is protected even if the borrower's collateral drops in price before they can react.

Also called: excess collateral · collateralized lending · over-margined borrowing

What it is

Overcollateralization is the standard risk-management pattern in onchain lending: to borrow an asset, you must deposit collateral worth more than the loan itself — commonly 120-150% of the loan value or higher. Because there's no credit score, court system, or ability to repossess a defaulting borrower's real-world assets, the protocol substitutes a buffer of extra value it can seize automatically instead.

How it works

  1. A borrower deposits an asset (ETH, a liquid staking token, another crypto asset) into a lending protocol's smart contract as collateral.
  2. The protocol assigns that collateral a loan-to-value (LTV) limit — say 75% — meaning the borrower can draw a loan worth at most 75% of the deposited collateral's current market value.
  3. The borrower takes out a loan (often a stablecoin or another crypto asset) up to that limit, leaving a cushion between the loan value and the collateral value.
  4. An oracle continuously feeds the protocol current market prices for the collateral asset.
  5. If the collateral's price falls (or the loan's value rises relative to it) enough that the position crosses a liquidation threshold — the point where the buffer has eroded to a minimum safety margin — the position becomes eligible for liquidation.
  6. Liquidators (often bots) repay some or all of the outstanding loan on the borrower's behalf and, in exchange, seize the borrower's collateral at a discount, pocketing the difference as a liquidation bonus/incentive.
  7. This process needs no court order or identity check — it executes automatically and immediately whenever the price crosses the threshold, protecting the protocol's solvency in real time.

Why designers use it

  • Enables trustless lending to pseudonymous, unvetted borrowers with no credit history or legal recourse, since the collateral buffer (not the borrower's promise) secures the loan.
  • Keeps the protocol solvent through volatile price swings by giving automated liquidators a financial incentive to close underwater positions before they go fully negative.
  • Removes the need for human underwriting, courts, or debt collection — everything is enforced by code and market incentives.
  • Lets the same collateral simultaneously back a loan while (in some designs) still earning yield or voting rights, improving capital efficiency for the borrower relative to simply selling the asset.

Failure modes

  • Cascading liquidations: a sharp price drop can trigger many liquidations simultaneously, dumping seized collateral onto thin markets, driving the price down further and triggering yet more liquidations — a self-reinforcing spiral.
  • Oracle failure or manipulation: if the price feed is stale, wrong, or manipulated (e.g., via a flash loan attack on a thin-liquidity price source), positions can be liquidated unfairly or, conversely, unsafe positions can avoid liquidation until it's too late.
  • Bad debt: if collateral value falls faster than liquidators can act (gas spikes, network congestion, extreme volatility), a position's debt can end up exceeding its collateral value, leaving the protocol with a loss it must absorb or socialize.
  • Capital inefficiency: locking up 150% or more of a loan's value ties up capital that could otherwise be productive, making overcollateralized lending inherently less efficient than undercollateralized (credit-based) alternatives.
  • Reflexive collateral: using a protocol's own governance or liquidity token as collateral creates circular risk, since a crash in that token can simultaneously wipe out both the collateral value and the protocol's broader health.

What to check before using it

  • What price oracle feeds the liquidation logic, how manipulation-resistant is it, and what happens if it goes stale or is attacked?
  • Are liquidation thresholds and bonuses calibrated to realistic worst-case volatility and network congestion, not just typical conditions?
  • Is there a mechanism (insurance fund, socialized losses, backstop auction) for the scenario where liquidations can't keep up and bad debt accrues?
  • How correlated is the collateral asset with the borrowed asset or with the protocol's own token, and what happens during a correlated crash?
  • Is collateral liquidity deep enough that a large liquidation won't itself crash the market it's selling into?

Experiments that used it · 2

Shown oldest first, so you can watch the design evolve.

Compound (lending)
Algorithmic pooled money market on Ethereum whose cToken model and June 2020 COMP liquidity-mining launch set the template for DeFi lending and ignited 'DeFi Summer'.
2018 major success
Falcon Finance
DWF Labs-incubated 'universal collateralization' protocol that mints an overcollateralized synthetic dollar (USDf) against crypto, stablecoin, and tokenized RWA collateral, then survived a July 2025 depeg scare with a public transparency push.
2025 ongoing