Skip to content
Financial Context Wiki

Liquidation Cascade

A liquidation cascade is a feedback loop in which forced position closures move the market enough to trigger further liquidations. Learn how derivatives and DeFi cascades form, how to read the data, and where the main risks lie.

Updated

For educational purposes only; not investment advice. Leveraged trading can result in rapid losses.

Direct answer

A liquidation cascade is a feedback loop in which forced position closures move the market enough to make additional accounts eligible for liquidation. In a downward cascade, liquidated longs create sell-side pressure; in an upward cascade, liquidated shorts create buy-side pressure. Thin liquidity, crowded leverage, correlated collateral, and rapid repricing make the loop more severe.

Liquidation is the risk-control event for one account or position. A cascade is the market-level sequence that can follow when many liquidation thresholds are close together and forced execution affects the prices used by other accounts. It is a descriptive market-structure term, not a standardized statistic or a guarantee that every sharp move was caused by liquidations.

Centralized derivatives venues and on-chain lending protocols implement liquidation differently. A venue may cancel orders, partially reduce a position, take it over, use liquidity providers, draw on an insurance fund, or invoke auto-deleveraging (ADL). A lending protocol may let a third-party liquidator repay eligible debt and seize collateral with a protocol-defined bonus. Always read the rules for the specific venue, account mode, contract, collateral asset, oracle, and protocol version.

How it works

The common feedback sequence is:

  1. An initial price move reduces the equity of leveraged accounts or the health factor of collateralized loans.
  2. A risk reference reaches a threshold. Many derivatives venues use a mark price rather than the last traded price; Aave makes a borrow position eligible for liquidation when its health factor falls below 1.
  3. Forced risk reduction reaches the market. Closing a long requires selling, while closing a short requires buying. The trigger price and the eventual execution price can differ, so slippage matters.
  4. Those trades move through available liquidity. If their price impact changes mark prices, index inputs, or oracle prices enough, another cluster of positions crosses its threshold.
  5. The process repeats until price stabilizes, vulnerable positions are exhausted, new liquidity arrives, or a venue’s risk controls interrupt the path.

Risk backstops manage losses but do not necessarily stop price feedback. BitMEX’s rules allocate liquidation-trade profit or loss to its insurance fund up to the amount assigned to the contract and provide for ADL under its rules. Coinbase documents a venue-specific waterfall involving order cancellation, position sales, collateral sales, liquidity support, an insurance fund, ADL, and clawbacks. These examples must not be generalized into one universal sequence.

On Aave, the health factor is (total collateral value × weighted average liquidation threshold) / total borrow value. An eligible liquidator repays debt and receives collateral plus a liquidation bonus. The seized collateral may be held rather than sold, so an on-chain liquidation does not automatically equal immediate spot selling. A cascade becomes more likely when liquidators repeatedly sell the same collateral, oracle updates reprice many accounts together, or one collateral asset backs positions across several protocols.

Worked example

Consider a simplified downward cascade. Ignore fees, funding, cross-margin effects, and venue-specific tiers; these assumptions make the arithmetic illustrative rather than a liquidation-price formula.

  • A cluster of longs totaling 1,000 BTC becomes eligible near a mark price of 57,000 USDT.
  • The liquidation engine’s sales consume bids and average 56,700 USDT. That execution is not the same thing as the mark-price trigger.
  • The market impact pushes the mark price through 56,500 USDT, where another 750 BTC of longs becomes eligible.
  • The second wave trades into thinner bids, and the market reaches 55,800 USDT before fresh buyers absorb the flow.

Across the two waves, 1,750 BTC became eligible for forced reduction. The move from 57,100 USDT immediately before the first threshold to 55,800 USDT is (57,100 - 55,800) / 57,100 = 2.28%. That does not prove the liquidations alone caused the entire move: active sellers, hedges, arbitrage, and changes on other venues may have contributed.

Nor does 1,750 BTC necessarily mean 1,750 BTC of net spot selling. Positions may be partially liquidated, transferred to liquidity providers, internally matched, hedged elsewhere, or closed using venue-specific procedures. Public liquidation feeds often cover only selected venues and may report notional value rather than executed market flow.

Risks and monitoring

Assess cascade risk in four layers:

  • Position susceptibility: Compare open interest with recent volume and credible depth; review leverage limits, maintenance-margin tiers, funding, basis, account mode, and the collateral used for margin. Funding and open interest can show crowding, but neither reveals a complete map of liquidation prices.
  • Trigger design: Identify the mark-price or index methodology for derivatives and the oracle, liquidation threshold, close factor, and update behavior for lending protocols. A chart’s last price may not be the liquidation trigger.
  • Execution capacity: Examine bid and ask depth, spread, expected slippage, concentration across venues, and whether liquidity providers are likely to reduce capacity during stress. Displayed orders can be cancelled and are not guaranteed absorption.
  • Loss containment: Read the rules for partial liquidation, insurance funds, ADL, clawbacks, protocol reserves, and bad debt. A backstop protects a particular part of the system; it does not guarantee uninterrupted trading, withdrawals, or user recovery.

For account-level risk, do not make survival depend on a rebound or on a stop order executing first. Reduce position size, maintain a margin buffer, understand whether cross margin exposes unrelated positions, and pre-plan exits while markets are orderly. Bybit states that liquidation can occur when the mark price reaches the threshold even if a stop keyed to the last traded price has not triggered.

Spot-only holders can still be exposed indirectly. Collateral sales can pressure spot markets, a falling margin asset can weaken several positions at once, and impaired stablecoin or wrapped-asset liquidity can connect derivatives and DeFi venues. Stress tests should vary liquidity, correlations, collateral haircuts, and execution slippage, not price alone.

Common misconceptions

Myth 1: Reported liquidation notional equals net market selling

Notional statistics do not show how much was partially reduced, internally matched, assigned, hedged, or executed against the public order book. Upward cascades also involve forced buying rather than selling.

Myth 2: Falling open interest proves the cascade is over

Open interest can fall because positions were closed, transferred, expired, or revalued in a reporting currency. Remaining leverage, on-chain debt, options hedges, and continuing discretionary sales can keep risk elevated.

Myth 3: An insurance fund prevents a liquidation cascade

An insurance fund is a loss-allocation layer after or during liquidation under venue-specific rules. It does not preserve the liquidated position and does not create order-book depth that guarantees orderly execution.

Myth 4: Every liquidation creates an immediate spot trade

Derivatives positions can be assigned or offset within a venue, and an on-chain liquidator can retain seized collateral. Spot pressure is possible and often important, but the mapping is not one-for-one.

Sources

Navigation

Search the wiki...