Quant Memo
Core

Liquidations and Cascade Risk

When a leveraged crypto position's collateral falls below what its loan requires, it gets forcibly closed — and because that forced selling pushes price down further, one liquidation can trigger the next.

Prerequisites: DeFi Lending and Utilization Curves

Borrow $8,000 in stablecoins against $10,000 of a volatile token as collateral, and the loan is fine as long as the collateral stays comfortably above the debt. Let the token's price fall and that cushion shrinks. Cross a threshold set by the protocol, and the position gets forcibly closed — sold, whether the borrower agrees or not. That forced sale is a liquidation, and it is the mechanism that keeps leveraged crypto lending solvent without a bank's ability to just call the borrower and ask for more collateral.

A position is liquidated when collateral value falls to a fixed fraction of debt, called the liquidation threshold. The liquidation itself is a forced sale into the open market — and because that sale pushes price down, it can trigger the next position's liquidation, turning one bad trade into a cascade.

How the trigger works

Every collateralized loan has a health factor, roughly the ratio of collateral value to debt, scaled by how conservative the protocol wants to be:

H=collateral value×liquidation thresholddebt valueH = \frac{\text{collateral value} \times \text{liquidation threshold}}{\text{debt value}}

In words: take what the collateral is worth, discount it by a safety margin the protocol sets (say, only counting 80% of its value), and compare that to what's owed. As long as H>1H > 1, the position is safe. The moment HH drops to 1, a liquidator — anyone, usually a bot — can step in, repay part of the debt, and seize the discounted collateral as a reward, often called a liquidation bonus.

price falls position liquidated forced sale the forced sale pushes price down again, feeding the loop
Each liquidation's forced selling can push price into the next position's threshold, so a single sharp move can unwind into a self-reinforcing cascade.

Worked example

A trader deposits $10,000 of ETH as collateral and borrows $7,000 in stablecoins, at a protocol liquidation threshold of 80%. Their health factor starts at H=(10,000×0.80)/7,000=1.14H = (10{,}000 \times 0.80) / 7{,}000 = 1.14. If ETH drops 15%, collateral value falls to $8,500, and H=(8,500×0.80)/7,000=0.97H = (8{,}500 \times 0.80)/7{,}000 = 0.97 — below 1, so the position is now eligible for liquidation. A liquidator repays some of the $7,000 debt and receives ETH collateral worth that amount plus a bonus (commonly 5–10%), meaning the borrower loses both their collateral and pays an effective penalty for having been under-margined.

What this means in practice

Liquidations are individually rational — the protocol needs someone incentivized to act fast — but collectively they're what makes crypto crashes sharper than the initial news would suggest. A 10% drop can force enough leveraged positions to sell simultaneously that it becomes a 20% drop, which triggers the next band of thresholds, and so on. This is the same mechanical logic as a margin call cascade in traditional markets, compressed into minutes instead of days because liquidation bots operate continuously and settlement is instant.

Health factor is not a fixed cushion — it moves with price, and the closer a position sits to H=1H = 1, the less price movement it takes to wipe it out. Traders often underestimate how a modest, ordinary-looking drawdown can be amplified by everyone else's leverage unwinding at the same threshold, not just their own.

Related concepts

Practice in interviews

Further reading

  • Aave Protocol Documentation, 'Liquidations'
ShareTwitterLinkedIn