On October 10, 2025, over $19 billion in crypto leverage was liquidated in about a day, sending crypto prices to levels that are still considered a “tail risk” event. This crash marked the start of a broader sell-off in cryptocurrencies that continued through December. It illustrates how leverage, liquidity and venue design interact when markets are stressed.
From the price title to the margin spiral
The October 10 crash followed a pattern that has become all too familiar in 2025. An all-Chinese tariff threat hit global risk assets, with crypto suffering the harshest backlash. The fact that cryptocurrencies react faster and stronger to negative news is not unusual in itself: cryptocurrencies trade 24/7, and do not benefit from automatic trading halts or interventions commonly seen in other markets. What turned this selloff into such a crisis was the intersection of three factors: how risk was positioned, how it was financed, and how debt and infrastructure interacted under pressure.
Unlike Terra/Luna or FTX, this crash is not due to fraud, the insolvency or bankruptcy of a single large institution. In early October, open interest on BTC and ETH perpetual futures was high. Funding rates rose from around 10% annualized to nearly 30% on October 6, driven by Ethereum’s rally. Much of this exposure was focused on sites using unified (multi-asset) margining. Unified margining can be very effective in calm markets: profits from one position offset losses elsewhere, allowing traders to increase the overall size of their books while using capital more efficiently. However, in times of crisis, the same design ties portfolios to their weakest assets. These assets are generally long positions only, and sales of these assets may not be offset by profits generated from the short positions.
When trading account equity falls below thresholds, exchanges can force liquidation of positions before owners can react. Additionally, issues in crypto infrastructure, such as frozen exchange interfaces, have prevented some traders from dynamically managing their risks or moving capital between exchanges. The combination of high leverage and the exchanges’ automatic deleveraging (“ADL”) mechanisms turned this selloff into a margin-driven liquidation spiral.
Figure: Timeline of the October 10 stunt
How Liquidity Really Failed
Intraday order book data showed that BTC’s top book depth declined by more than 90% at key venues that day, with bid-ask spreads widening from single-digit basis points to double-digit percentages at the extremes. Liquidity still existed, but only in small quantities and at prices that did little to bridge the imbalance between sellers and buyers. As the size of executable orders evaporated, many market makers either widened their spreads significantly or withdrew altogether.
The October 10 crash reminded us of market lessons that are often forgotten during lulls and bull cycles: markets trade at the margin, not the average. Market depth is a non-linear variable that decreases sharply in times of crisis. Market impact increases when trade size remains constant and volume decreases. These dynamics exacerbate any sell-off through negative feedback loops. When some crypto books actually operate with 20-50x leverage, their main risk lies not in directional bets but in scenarios in which liquidity disappears and exchange infrastructures become unreliable at the same time.
When a Stablecoin lost its anchor
The most telling episode of microstructure was not the price action of major coins like BTC, but what happened to USDe, a delta-neutral stablecoin designed to hold 1:1 value with the US dollar. At the height of the sell-off, USDe traded in the mid-$0.60s on Binance – implying a roughly 35% discount – while on other exchanges and in several DeFi pools it remained much closer to $1.
This price divergence had little to do with fundamentals and everything to do with the microstructure of the sites. The key question was how local pricing powered the oracles and margin engines. Many leveraged products price collateral using the site’s spot price, either directly or through a simple internal oracle. When USDe collapsed on Binance, margin systems sharply reduced it, reduced the value of collateral, and pushed thousands of accounts beyond maintenance thresholds. Positions that could have remained solvent under multi-site pricing were liquidated simply because the Binance market traded via pegging.
When the backstop must liquidate the winners
Stock exchanges protect themselves from leveraged traders whose accounts go into deficit when markets go sideways. When standard tools such as margin calls, liquidations, and support programs (if applicable) are not enough, the ADL steps in, forcibly closing profitable positions on the other side of these trades to make up the shortfall. This creates a distinct risk: ADL may unintentionally close profitable short positions, thereby transforming a covered portfolio into a naked portfolio during periods of stress.
On October 10, some of the best-hedged shorts had their positions reduced or closed entirely to maintain FX solvency. From an exchange’s perspective, ADL is preferable to outright default or a retrospective socialized loss (“clawback”). From the trader’s perspective, this adds a second layer of risk: the underlying economics of the trade and the rules that dictate whether it will be allowed to persist when it matters most. Offices that rely on derivatives as hedging tools should incorporate ADL mechanisms into their counterparty due diligence.
The structural gap compared to traditional markets
Traditional markets built safety guardrails after their own flash crash episodes: circuit breakers, central counterparties with conservative margin models, and strict debt limits for retail trading. Crypto platforms play many of these roles at once, but risk is concentrated in platform-level margin drivers, high leverage is widely available, and price discovery is fragmented across exchanges with inconsistent practices. The October crash exposed the flaws in this framework, particularly where high leverage, exotic collateral and internalized pricing are at play.
What has changed and what has not changed
Two months later, some surpluses have been drained from crypto markets. Crypto markets have weakened, deviating from the broader trend in most other risk assets. BTC has fallen around 30% from its all-time high in early October. Bitcoin ETFs have seen significant outflows. Open interest is down more than 40% from October highs.
Funding levels for perpetual futures contracts have normalized. Several major venues have tightened leverage caps on certain pairs, increased haircuts on flimsy collateral, and announced moves toward multi-venue pricing for key oracles. Millions of accounts were closed, leveraged players were driven out, and retail flows into leveraged products cooled.
However, the underlying economic principles remain the same. As long as traders are willing to pay for leverage, someone will offer it. This keeps the cycle going and ensures that crypto remains a volatile asset class.
Lessons for the next cycle
For traders and risk managers, overall leverage is just a starting point. The effectiveness of leverage depends on the behavior of the guarantee basket in a crisis situation. Risk management teams should model executable sizes, not just spreads, and build scenarios in which market depth decreases by 90% over short periods of time. Concentrating positions, collateral, and oracles in a single location saves margin in calm periods, but can turn into a single point of failure when conditions become unstable.
For exchanges and infrastructure providers, the priority is “plumbing first.” Multi-site, liquidity-weighted oracles with outlier controls are expected to become the norm. Liquidation drivers should be tested against brief, violent upheavals rather than gentler historical trajectories. Transparent documentation of margin logic, haircuts and ADL triggers would benefit all participants.
For allocators, due diligence now extends to how platforms handled this episode: venues, rules, and risk infrastructure belong on the same terms as counterparties and legal terms.
For highly exposed participants, these questions are no longer optional.
