Liquidations close undercollateralized loans by selling enough collateral to repay the debt and compensate the liquidator. That single mechanism lets an on-chain lending market recover value without waiting for a borrower to return, negotiate, or appear in court.
What has to be true first
A liquidation becomes possible when the value of the collateral, after the protocol’s haircut, no longer supports the outstanding debt. In an Aave-style market, the key calculation is the health factor: collateral value multiplied by its liquidation threshold, divided by the debt value. Below one, the account is eligible for liquidation.
The threshold is deliberately below the maximum loan-to-value ratio. If a market allows borrowing up to 80 percent of collateral value, it may begin liquidation at a lower effective safety margin. That gap gives liquidators room to sell collateral before a fast price move turns a recoverable loan into bad debt.
The protocol does not use the borrower’s wallet balance or intentions to make this decision. It uses debt that has accrued interest, collateral priced by an oracle, and the relevant market parameters. A borrower can be liquidated while still holding plenty of unrelated assets elsewhere; the protocol can act only on collateral pledged to that position.
What the liquidation transaction produces
A liquidator repays some or all of the borrower’s debt and receives collateral at a discount called the liquidation bonus. If a borrower owes 10,000 units of a stablecoin and the bonus is 7 percent, repaying that debt may entitle the liquidator to collateral worth 10,700 units at the protocol’s quoted price. The borrower loses the collateral, the debt is reduced, and the market receives a buyer precisely when the position is under stress.
Most protocols limit how much debt can be repaid in one transaction. That close factor prevents a liquidator from taking the entire position when a partial repayment is enough to restore solvency. Other designs use auctions, especially when the collateral is less liquid or the position is too large for an instant market sale.
The bonus is not free yield. It pays for gas, execution risk, slippage, and the possibility that the collateral price moves between the oracle update and the sale. If the bonus is too small, liquidators ignore marginal accounts. If it is too large, healthy borrowers pay more for the protection and liquidations can create unnecessary selling pressure.
Why the mechanism matters during a price move
Liquidations turn a falling collateral price into a forced reduction of leverage. Suppose a borrower deposits 20 ETH and borrows stablecoins against it. If ETH falls sharply, the position crosses its liquidation threshold. A liquidator repays debt, takes part of the ETH, and leaves the borrower with less collateral and less debt. The protocol has converted a bad position into a smaller one before the loss reaches depositors.
That is the problem automated lending markets remove. Before this mechanism, a lender facing a default had to contact the borrower, seize assets through a trusted intermediary, or rely on legal enforcement. Smart contracts cannot make collateral valuable, but they can make the claim on that collateral executable under predefined conditions.
The important risk is not merely that liquidation happens. It is that collateral may be impossible to sell at the oracle price. Thin liquidity, a delayed oracle, a congested chain, or a rapidly falling asset can leave the protocol with less value than the debt. The liquidation bonus helps attract buyers; it cannot eliminate market impact. The remaining shortfall becomes bad debt, absorbed according to that protocol’s reserve and loss-sharing design.
When liquidation is the wrong tool
Liquidation is for a position that has breached its risk boundary, not for every borrower who wants to reduce exposure. A borrower who still has room should usually repay voluntarily, add collateral, refinance, or swap into a less volatile asset. Those actions avoid the bonus, the forced sale, and the possibility of receiving a poor execution price.
It also does not replace credit underwriting for unsecured or undercollateralized lending. If repayment depends mainly on a borrower’s income, reputation, or legal promise, there may be no liquid asset for a smart contract to sell. That is a different lending model with different enforcement costs.
Cross-chain movement is another separate operation. A position on Manta Pacific remains governed by its lending market’s debt and oracle state; moving an asset across Manta Network does not repay that debt. Celer Network may be relevant to the transfer leg, but a bridge transaction does not by itself improve a loan’s health factor. When the choice is between unwinding the position and moving assets through Manta Bridge, the cross-chain route is the separate path to inspect: graph.org