Crypto markets are often described through prices: the percentage bitcoin lost, the depth of an altcoin decline or the value of positions liquidated. Those figures record the outcome. They do not, by themselves, explain the mechanism.

The more revealing question is what happens between the first price movement and the eventual market low. Leveraged positions breach maintenance requirements, collateral loses recognised value, market makers reduce risk, order books thin and prices diverge across venues. Stablecoins, which are widely used for trading and settlement, can themselves trade away from par or become difficult to redeem. Automated liquidations then convert falling prices into compulsory transactions.

This is market plumbing: the arrangements through which orders, collateral, funding and settlement interact. Stress exposes where those arrangements are robust and where apparently abundant liquidity depends on favourable prices, functioning infrastructure and intermediaries continuing to quote.

The distinction between evidence and inference matters. Public data can show that open interest fell, liquidations occurred or a token traded below its reference value. It may not reveal every participant's leverage, internal credit limits or the sequence of trades across bilateral markets. A defensible account therefore separates recorded facts from conclusions suggested by market mechanics.

The October 2025 cascade was larger than its trigger

On 10 October 2025, crypto prices fell sharply over about half an hour before partially recovering. The Financial Stability Institute's April 2026 paper described the episode as the largest cascade of crypto liquidations observed to that date. It reported direct liquidation losses of $19 billion on 11 October and expressly characterised that estimate as a lower bound because exchanges restrict how many liquidation events they report each second.

Coinbase Institutional reported that the event reduced its systemic leverage ratio, calculated as derivatives open interest divided by crypto market capitalisation excluding stablecoins, from approximately 7 per cent to below 4 per cent. This is a proprietary indicator rather than a complete measure of market leverage. Open interest can include hedged positions, and the ratio does not reveal individual balance sheets.

The evidence nevertheless fits a recurring mechanism. Falling prices reduce the equity supporting leveraged long positions. Exchanges and protocols then close positions that no longer meet maintenance requirements. Those liquidations create sales or auctions in a market already declining. If bids are limited, execution moves to progressively lower prices, weakening other accounts and triggering further liquidations.

The FSI paper identified high volatility, low liquidity in some cryptoassets, automated derivatives liquidation, high available leverage and the use of cryptoassets as margin as likely amplifiers. It did not claim that leverage caused the initial shock. The supported conclusion is narrower: leverage and automated risk controls intensified the subsequent movement.

Why visible liquidity can disappear

Liquidity is not synonymous with trading volume. A market can report substantial turnover while remaining unable to absorb concentrated, urgent selling without a large price change. During stress, the relevant properties are executable depth, bid-offer spreads, the replenishment of quotations and the capacity to move inventory between venues.

Much displayed liquidity is conditional. Market makers quote across venues while managing inventory, exchange credit, margin and hedges. When volatility rises, they face a greater risk that a quotation will be executed just before the wider market moves against them. They may widen spreads, reduce order sizes or withdraw quotes. Each action reduces the depth available to leveraged sellers.

Coinbase's 28 October 2025 market review said many altcoins fell by 40 to 70 per cent during the 10 October episode and described market makers pulling quotes. Coinbase also attributed the cascade partly to fragmented altcoin liquidity and automatic unwinding by decentralised protocols. These are the assessments of a market participant, not an independent regulatory reconstruction, but they are consistent with the FSI paper's finding that thin liquidity amplified the event.

Order-book depth is path-dependent. A sale consumes the best bids first. Unless liquidity providers replace them, the next sale executes at a worse price. A liquidation engine designed to reduce exposure quickly may prioritise execution over price. Several engines acting together can overwhelm the quantity available near the previous market price.

Stress does not merely reveal how much liquidity was displayed. It reveals how much remained executable after volatility, credit constraints and operational risk changed providers' incentives.

This explains why volume can rise while market quality deteriorates. High turnover may reflect rapid trading across a widening range of prices. It does not prove that participants could transact at stable prices.

Leverage creates a mechanical feedback loop

Leverage allows a participant to control an exposure larger than the capital committed to it. Perpetual futures, margined futures, lending protocols and collateralised borrowing have different legal and technical structures, but they share a similar stress logic.

  1. A decline reduces the equity supporting a leveraged long position.
  2. The venue recalculates the account against its maintenance requirement.
  3. If margin is insufficient, the trader must add collateral, reduce the position or face liquidation.
  4. Liquidation adds selling pressure to a falling market.
  5. The new price weakens other accounts and may reduce the value of collateral supporting unrelated positions.

This feedback is procyclical. Rising prices increase collateral values and can support more borrowing. Falling prices reverse the process, often more abruptly because automated systems impose fixed thresholds.

Liquidation can protect a venue or lender by closing an account before its equity becomes negative. That function should not be confused with price stability. A venue can remain solvent while its liquidation process contributes to disorderly execution. Partial liquidation, position limits, insurance funds and staged auctions may moderate the process, but their effectiveness depends on calibration and available counterparties.

Collateral can fail when it is most needed

Leverage is particularly fragile when the financed asset and its collateral are exposed to the same shock. If a trader holds a leveraged crypto position and posts another cryptoasset as margin, a broad decline damages both sides of the account.

Haircuts recognise this risk by crediting collateral at less than its market value. A haircut calibrated to normal volatility and liquidity may still prove inadequate during a discontinuous move. Correlations can also rise during stress, making an apparently diversified basket behave like a concentrated position.

Binance's 12 October 2025 incident statement said some platform modules experienced brief technical problems after 21:18 UTC on 10 October, while its core spot and futures matching engines and API trading remained operational. Binance said the wider sell-off reached its low between 21:20 and 21:21 UTC and that severe depegging in USDe, BNSOL and WBETH occurred after 21:36 UTC.

A separate Binance compensation notice covered affected futures, margin and loan users who held those assets as collateral during the depeg window from 21:36 to 22:16 UTC. Binance also said it would add redemption prices to relevant indices and introduce a minimum threshold for the USDe index.

These are Binance's own findings and compensation terms, not an independent audit. The FSI paper presented a less favourable account, reporting that a high-service-load outage restricted some investors from closing positions and produced pricing discrepancies between Binance and the wider market. The two accounts support the existence of operational and collateral-pricing problems but do not independently establish every part of the causal sequence.

The episode illustrates the difference between an asset's underlying claim and the price recognised by a risk engine at a particular moment. A redeemable or staked asset may retain a claim on underlying value while trading at a discount in a thin local market. If collateral is marked to that price, the discount can trigger liquidation and deepen the dislocation.

Fragmented venues turn one market into many

Crypto trades across centralised exchanges, decentralised exchanges, derivatives platforms, lending protocols and bilateral desks. There is no single consolidated order book or uniform margin system. Assets may also exist through wrappers and bridges with distinct operational and counterparty risks.

An ESMA analysis dated 10 April 2024, based on granular trading and order-book data supplied by Kaiko, described crypto secondary markets as global and complex. It found significant concentration in assets and exchanges, while liquidity varied considerably between exchanges and currencies. The data predate the October 2025 event, so the analysis supports the structural discussion rather than reconstructing that episode.

Fragmentation can transmit stress in several ways:

  • Prices can diverge because venues have different participants, collateral rules and depth.
  • Arbitrage requires capital on the correct venue and the ability to transfer assets promptly.
  • Congestion, withdrawal restrictions or outages can delay those transfers.
  • Liquidation engines can use different indices, mark prices or local prices.
  • A margin call on one venue may not be met with collateral held elsewhere.

Under ordinary conditions, arbitrage tends to keep markets aligned. During stress, the balance sheets performing that arbitrage may become constrained. A price difference is not risk-free profit if the asset cannot be transferred, financed or redeemed before the gap changes.

Fragmentation can therefore amplify temporary dislocations, but its precise contribution to any event requires multi-venue transaction data. Public dashboards and exchange statements cannot fully reconstruct internal credit lines, rejected orders, latency or off-exchange hedges.

Stablecoins are liquidity instruments

Stablecoins connect crypto markets to fiat money and serve as quote assets, collateral and settlement instruments. Stress in a stablecoin can therefore affect the unit in which positions are margined and the means by which traders move purchasing power between venues.

The central distinction is between primary convertibility and secondary-market price. Eligible customers may redeem a fiat-backed stablecoin with its issuer. Other holders trade on exchanges or through automated market makers. If direct redemption is unavailable or delayed, secondary sellers must find buyers and may accept a discount.

The March 2023 USDC episode provides detailed evidence. A Federal Reserve analysis published on 17 December 2025 records that Circle announced on 10 March 2023 that it could not withdraw $3.3 billion of USDC reserves from Silicon Valley Bank, around 8 per cent of total reserves. Primary redemptions rose and then became minimal over the weekend as banking hours constrained operations. USDC reached 86 cents on secondary markets, while hourly trading volume approached $2 billion on 11 March.

The study found that decentralised exchanges accounted for most activity during the most intense period, although centralised exchanges normally handled more crypto trading. It also documented contagion through Dai's one-for-one exchange facilities with USDC and other stablecoins. Those facilities were rapidly drained, while Dai and stablecoins without direct exposure to Silicon Valley Bank also moved away from par.

On 12 March 2023, the US Treasury, Federal Reserve Board and Federal Deposit Insurance Corporation announced that Silicon Valley Bank's depositors would be fully protected. USDC recovered sharply following the announcement and returned to par after Circle resumed redemptions on 13 March. The Federal Reserve study also records Circle's 15 March statement that it had cleared substantially all of its minting and redemption backlog.

The evidence shows that reserve quality alone does not eliminate liquidity risk. Access, operating hours and settlement capacity also matter. A stablecoin can have reserves sufficient on an eventual-value basis yet trade below par when immediate convertibility is uncertain.

Stablecoin reserves connect crypto and traditional markets

Reserve-backed stablecoins promise redemption while holding cash and securities. Redemption liabilities can be immediate, while reserve assets may require sale, transfer or maturity.

The Financial Stability Board's 16 February 2022 assessment identified liquidity mismatch, credit risk and operational risk as vulnerabilities that could make stablecoins susceptible to disruptive runs. It said disorderly reserve liquidation could spill over to short-term funding markets.

This is a risk channel, not proof that such a fire sale occurred in March 2023. The Federal Reserve's study treats forced Treasury sales as a counterfactual possibility. Public intervention protected bank deposits before that scenario could be observed.

The Basel Framework's cryptoasset standard shown as in force from 1 January 2026 applies to banks' cryptoasset exposures. For a stablecoin exposure to qualify for the framework's Group 1b treatment, its reserve assets must meet specified conditions, including reliable liquidity under stress, sufficient daily liquidity for immediate redemption requests and bankruptcy-remoteness protections. The framework excludes certain financing transactions from eligible reserves because they can increase leverage or temporarily substitute higher-quality assets.

These requirements should not be described as direct regulation of every stablecoin issuer. They are prudential classification conditions governing how banks treat qualifying exposures.

What the evidence establishes

  • Automated liquidations can concentrate compulsory transactions into short periods.
  • Crypto collateral can lose recognised value alongside the financed position.
  • Liquidity varies across assets and venues and may recede as volatility rises.
  • Fragmented custody, pricing and settlement can obstruct arbitrage.
  • Stablecoin secondary prices depend partly on confidence in prompt redemption.
  • Protocol and stablecoin interconnections can transmit stress beyond an original exposure.

Other claims require restraint. Public evidence does not identify one comprehensive cause for the October 2025 crash. Liquidation totals depend on reporting coverage. Venue statements reflect their authors' systems and incentives. On-chain records do not reveal all beneficial ownership, centralised positions or bilateral financing.

It would also be premature to describe crypto fragility as a demonstrated threat to the wider financial system. The evidence establishes plausible transmission channels, including stablecoin reserve holdings and growing links with regulated finance. It does not establish that systemic contagion occurred in the events examined here.

The enduring lesson is conditional liquidity

Round-the-clock trading and automated risk management create an impression of continuous liquidity. Stress reveals its limits. Trading can continue while banking rails are closed. Collateral can retain underlying value while becoming temporarily difficult to sell. A stablecoin can be backed while many holders cannot redeem immediately. An exchange can keep its matching engines running while other modules or transfer routes are impaired.

Leverage binds these weaknesses together. Margin calls may not wait for arbitrage, redemption or settlement. Collateral is marked at current recognised prices, not at an expected future recovery value. Reduced depth becomes further volatility because positions must be closed regardless of execution quality.

The useful variables are therefore executable depth, collateral quality, price-source design, redemption capacity, transfer resilience and the distribution of leverage. They determine whether an initial shock is absorbed, transmitted or mechanically amplified.

This is an analytical framework, not investment advice. Its central finding is that liquidity is a contingent service supplied by balance sheets and infrastructure, while leverage creates claims on that service precisely when it is most likely to become scarce.