How crypto liquidation cascades work, and why one forced sale triggers the next

A liquidation cascade is what happens when a forced sale of one leveraged position pushes the price down far enough to force the sale of the next one, and the next, in a self-feeding chain. It is not panic selling and it is not a coordinated attack – it is a mechanical consequence of leverage, thin order books and automated margin engines, the same underlying dynamic whether it plays out on a centralized exchange’s perpetual futures desk or inside a DeFi lending protocol.

The mechanics of one liquidation

Leveraged trading lets a trader control a position larger than their own capital by borrowing the difference. Startup Fortune’s worked example: a trader who puts up $1,000 in collateral and borrows another $4,000 controls a $5,000 position – 5x leverage. That trader needs the price to fall only 20% before the position is worth less than the borrowed amount. At that point, per Startup Fortune, the exchange does not ask permission. It sells the position automatically to recover its loan. Bitcoin.com News describes the same mechanism at larger scale: a trader can control $100,000 of bitcoin exposure with $10,000 of margin through a perpetual futures contract, a derivative that lets someone trade price exposure without owning the underlying coin. Once the price move erodes the margin below a maintenance threshold, the exchange closes the position into the open market, whatever the price on offer.

Why one liquidation becomes a cascade

One forced sale barely moves the price. The cascade starts when many leveraged positions sit at different price levels and the market begins moving through them in sequence. Bit’s Knowledge Hub lays out the mechanism in layers: an initial decline of around 5% liquidates the most heavily leveraged positions, those using 50 times leverage or more. Their forced selling drives an additional 2% decline, which is enough to liquidate the next tier, positions carrying 20 times leverage. A further 3% drop then reaches the 10-times-leverage tier, and so on until the leverage in the system is exhausted or buyers step in with enough size to absorb it. Bit calls the common misconception directly: liquidations do not all happen at once to everyone at the same level – they move through the market in stepwise “flushes” as price slices through clustered liquidation levels.

Cross-asset or unified margin accounts widen the chain further. Startup Fortune describes how, because many platforms let all of a trader’s positions share one collateral pool, a loss on one position eats into the margin backing another, so a Bitcoin cascade spreads sideways into Ethereum, Solana and other assets even when those assets have no fundamental link to Bitcoin. The correlation between otherwise unrelated tokens approaches 1 during a cascade, according to Startup Fortune, even for coins that move independently under normal conditions.

The same idea, a different rulebook: DeFi lending

Centralized exchange perpetuals are one venue for this mechanism. Decentralized lending protocols run a related but distinct version, according to Chainlink’s own documentation. Borrowers there must post collateral worth more than their loan – an overcollateralized structure defined by a loan-to-value, or LTV, ratio. If a protocol sets a maximum LTV of 75%, a borrower’s collateral must stay worth at least 25% more than the loan at all times, per Chainlink. Every open loan has a “health factor” measuring collateral value against the amount borrowed; once that health factor drops below one, the loan becomes eligible for liquidation. Automated bots called liquidators then repay the debt and seize the collateral, plus a penalty fee as their incentive, and immediately sell that collateral on a decentralized exchange to realize the profit. If the sale is large relative to available liquidity, it drives the asset’s price down, an oracle reports that new lower price back to the protocol, and other borrowers’ health factors drop below one in turn. Chainlink notes this is also where oracle design becomes a point of failure: if a price feed relies on a single low-liquidity exchange, or lags during network congestion, the protocol can liquidate loans based on manipulated or stale prices.

When liquidation alone isn’t enough: insurance funds and ADL

Centralized exchanges run insurance funds meant to absorb the gap when a liquidated position’s losses exceed its collateral. When the fund cannot keep up, exchanges turn to a further mechanism called auto-deleveraging, or ADL, according to FTI Consulting, PrimeXBT and Bitcoin.com News. ADL closes out profitable positions on the opposite side of a losing trade to cover the shortfall – meaning traders who did nothing wrong, and who may have been hedged, can have winning positions closed involuntarily to keep the exchange solvent. FTI Consulting frames this as a second layer of risk beyond the underlying trade: the rules governing whether a position is allowed to persist under stress.

October 10, 2025, worked through

The largest liquidation event on record, per CoinDesk Research, PrimeXBT and Bitcoin.com News, began after President Trump announced a 100% tariff on Chinese imports on October 10, 2025, a move also reported by Bloomberg, FTI Consulting and the other outlets above. Equities and commodities sold off alongside crypto, which was sitting near record open interest with crowded long positioning, according to CoinDesk Research.

FTI Consulting adds detail the others don’t: BTC and ETH perpetual funding rates, the periodic payments longs pay shorts, had climbed from around 10% annualized to nearly 30% by October 6, 2025, a sign positioning was already one-sided before the tariff news hit. A large share of that exposure sat on venues using unified, cross-asset margin, which FTI says is efficient in calm markets but ties an entire portfolio to its weakest asset once one position starts losing. FTI’s own intraday order-book data showed BTC’s top-of-book depth on key venues shrinking by more than 90% that day, with spreads widening from single digits in basis points to double-digit percentages at the extremes – liquidity did not disappear entirely, FTI writes, but what remained cleared almost nothing.

The stablecoin USDe, designed to hold parity with the dollar through a delta-neutral strategy, became what FTI Consulting calls the most revealing part of the episode. On Binance it traded far under $1 – FTI puts it in the mid-$0.60s, a roughly 35% discount; CoinDesk Research separately puts the low at $0.65. Neither outlet states whether these are the same measurement window, so this page reports both rather than choosing one. The mechanism FTI describes is the same oracle problem Chainlink documents for DeFi: many leveraged products priced collateral using the venue’s own spot price, so when USDe crashed specifically on Binance, margin systems there marked collateral down sharply and pushed accounts through liquidation thresholds that would have stayed solvent under a price blended across venues.

By December 24, 2025, FTI Consulting reported that open interest was still down more than 40% from the October highs, funding levels had normalized, and BTC had declined roughly 30% from its early-October all-time high – evidence, FTI writes, that the leverage built up before the crash had been substantially drained, even as the broader sell-off continued.

This is not the first time, and it wasn’t the last

Bit’s Knowledge Hub and Chainlink both point to March 12, 2020’s “Black Thursday,” when Bitcoin fell from around $8,000 to below $4,000 in under a day, per Bit, and Ethereum fell more than 50% in a single day, per Chainlink, triggering a wave of MakerDAO liquidations; network congestion delayed oracle price updates so badly that some liquidators won collateral auctions for zero cost, according to Chainlink. Total liquidations that day exceeded $1 billion, a record at the time, per Bit’s Knowledge Hub – a figure Bit notes is roughly a twentieth the size of October 10, 2025’s total. Startup Fortune separately traces the May 2022 collapse of Terra’s UST stablecoin and its sister token Luna, where Luna’s price fall from roughly $80 combined with a conventional leverage cascade to erase what CoinGecko’s data showed was a Terra ecosystem market capitalization once around $60 billion, and cites Coinglass data showing more than $8 billion liquidated in 24 hours during the separate May 19, 2021 crash. Startup Fortune also distinguishes that price-driven cascade from the counterparty cascade that followed FTX’s November 2022 collapse – when Alameda Research’s leverage against FTX’s own FTT token was exposed and Binance’s chief executive, Changpeng Zhao, announced he would sell Binance’s FTT holdings, triggering contagion into BlockFi, Voyager Digital and Genesis – which Startup Fortune says played out over years of bankruptcy proceedings rather than resolving within days. PrimeXBT and Bitcoin.com News report the pattern continued in 2026: more than 182,000 traders lost over $1.08 billion on January 20, 2026, mostly on long bitcoin and ethereum futures; roughly $2.2 billion was forcibly closed on February 1, 2026, an event traders dubbed “Black Sunday II,” with ethereum losses of $961 million and bitcoin losses of $679 million as bitcoin briefly broke below $76,000; and bitcoin’s slide from about $67,000 to $59,100 over 48 hours in June 2026 triggered over $3 billion in forced liquidations, including a single day near $1.8 billion.

What this page does not tell you

The headline $19 billion figure for October 10, 2025 comes from Coinglass, a data aggregator that compiles what exchanges choose to report through their own liquidation feeds – it is not an audited or regulator-verified total. PrimeXBT and Bitcoin.com News both report that unnamed market makers believe the real figure may have reached $30-40 billion, because some exchanges cap or delay what they disclose; this page cannot say which number is closer to the true total, and no named source in the evidence puts a figure on the gap.

FTI Consulting and CoinDesk Research give different numbers for how far USDe fell on Binance that day – mid-$0.60s versus $0.65 – and neither article states its measurement window precisely enough to reconcile the two. The claim that 1.6 million traders were affected appears only in PrimeXBT and Bitcoin.com News, not in CoinDesk’s or FTI’s accounts, and this page cannot independently verify it. FTI’s figure that top-of-book depth fell more than 90% is drawn from FTI’s own intraday order-book data and is not broken out by specific exchange in the evidence reviewed, so it cannot be checked against a named venue. Finally, this page describes how liquidation engines, insurance funds and auto-deleveraging generally work and how they behaved on October 10, 2025, but it does not reproduce any single exchange’s actual rulebook – margin tiers, ADL trigger thresholds and haircut schedules vary by venue and by asset, and, as FTI Consulting itself argues in calling for “transparent documentation of margin logic, haircuts and ADL triggers,” that documentation is not currently standard across the industry.

Sources

Every fact above is attributed to one of these reports. Where they disagree, the article says so.

• TheCoinrise reference desk
• TheCoinrise reference desk
• TheCoinrise reference desk
• TheCoinrise reference desk
• TheCoinrise reference desk
• TheCoinrise reference desk
• TheCoinrise reference desk
• TheCoinrise reference desk