The October 2025 Liquidation Cascade
Leveraged positions collapsed across the market in one night. What roles did prices, oracles, and automatic deleveraging play?
Checked in September 2026. This is event-layer content; read the final “Expiring Details” section first.
What Happened
| Date | Event |
|---|---|
| October 10–11, 2025 | After macroeconomic news, Crypto assets fell sharply within hours. Multiple data platforms reported the largest recorded 24-hour liquidation total, on the order of tens of billions of dollars |
| The same night | A stablecoin price used internally by a major exchange briefly diverged sharply. Some positions using it as collateral were liquidated; the exchange subsequently provided compensation |
| The same night | Some perpetual platforms triggered automatic deleveraging: profitable positions were forcibly closed to cover losing positions' shortfalls |
Analyze with Course Models
Why a Cascade Forms
The feedback loop from Chapter 16 and Chapter 18: falling prices → shrinking collateral → liquidation sales → further price declines. Each platform follows its own rules, but together they form Chapter 21's downward spiral. Greater leverage means a smaller initial decline can start the loop.
Who Supplies the Price?
T21 covers oracles: how chains learn outside prices. This night exposed that a platform's internal valuation price may differ from prices genuinely executable elsewhere. If valuation references only its own thin order book, one large sell order can make collateral appear to fall sharply, triggering otherwise unwarranted liquidations. This is Chapter 24's price-manipulation mechanism, even without deliberate manipulation.
Why Profitable Positions Also Close
T23 explains perpetual risk engines. When losing traders' margin cannot cover losses, insurance funds pay first. If they are insufficient, the system takes from profitable traders, ranking positions by profit and leverage and forcibly closing some. Chapter 18 already described this case; that night demonstrated it at extraordinary scale.
Retain the Mechanism
- Liquidation is a feedback loop: reasonable individual-platform rules cannot prevent every platform selling simultaneously.
- Valuation and executable prices differ: greater dependence on a single thin market increases oracle errors during extremes.
- Someone must cover the shortfall: margin → insurance fund → profitable traders. Chapter 23's risk engine fixes this order in rules.
The Eighteen Questions
First the position holder, then the insurance fund, and finally profitable traders whose positions are automatically reduced even though they did not make the losing leveraged trade.
Market makers cancel orders during declines, rapidly thinning depth. Liquidity is scarcest precisely when needed most (Chapter 15).
Verify Independently
Find a perpetual platform's official explanation of that night. Record its mark-price method, insurance-fund usage, and whether automatic deleveraging occurred.
Find that night's liquidation data on an onchain data platform. Specify source and measurement definition, following Chapter 23: define the metric before examining numbers.
Use Chapter 16's health formula to calculate whether a 10× leveraged position would have been liquidated at that night's largest decline.
Expiring Details
- Platforms use different definitions for totals; consult original data. This page deliberately avoids exact amounts.
- Platforms subsequently revised mark-price, insurance-fund, and automatic-deleveraging rules. Check current versions.