Market Microstructure | Derivatives | June 2026
The Liquidation Cascade Atlas: Every Major Crypto Exchange's Open Interest Map and the Price Levels That Will Wipe Out Billions in an Hour
Most crypto traders know liquidation cascades happen. Almost none know where the clusters sit before price reaches them, despite that information being derivable from public open interest and funding rate data across major exchanges. This article forensically maps the mechanics of liquidation cluster formation across Binance, Bybit, OKX, Hyperliquid, and dYdX, and walks through four of the largest cascades in crypto history with verified figures: the May 19, 2021 "Black Wednesday" crash, in which more than $8 billion in leveraged positions were liquidated within 24 hours across over 800,000 trader accounts as Bitcoin fell roughly 30% from $43,000 to $30,000 in approximately 12 hours; the November 2022 FTX collapse, in which Bitcoin fell from roughly $20,500 to $15,600 and the broader crypto market lost approximately $236.7 billion in two days; the August 2024 Japanese yen carry-trade unwind, which produced one of Bitcoin's sharpest single-week declines of that year; and the October 10, 2025 cascade, the largest in crypto history at $19.13 billion liquidated within 24 hours across 1.6 million trader accounts. The centerpiece instrument, the DN Liquidation Atlas, models the current distribution of open interest across major exchanges and the price-band clusters where the largest dollar value of forced liquidations would trigger, producing a live-adjustable cascade risk score that rises as open interest concentration increases near the current price.
Every trader who has been liquidated remembers the price level. Almost none of them knew, before it happened, that the level was visible in advance. Open interest data, funding rates, and order book depth are all public. The price bands where the largest dollar value of leveraged positions sit, waiting to be forcibly closed, are not secret information available only to exchanges and market makers. They are simply information most retail traders have never been shown how to read.
This article does two things. First, it forensically walks through four of the largest liquidation cascades in crypto history, with verified figures, to establish exactly how a cascade mechanically forms and accelerates. Second, it gives you a live, adjustable model for reading where the current clusters sit relative to today's price, so you can position around the cascade rather than inside it.
A liquidation cascade does not start with a sell button. It starts with a leveraged trader's collateral dipping below maintenance margin, an exchange's automated liquidation engine closing the position via forced market sell regardless of price impact, and the resulting sell pressure pushing price into the next leverage tier's threshold. The entire sequence is mechanical, not discretionary, which is exactly why it is mappable in advance.
— Synthesis of documented liquidation engine mechanics across major centralised exchanges.Four Cascades, Forensically Verified
May 19, 2021 — "Black Wednesday"
Bitcoin fell approximately 30% from roughly $43,000 to $30,000 within about 12 hours, triggered by a combination of Beijing reiterating its crackdown on crypto transactions and Tesla's announcement days earlier that it would no longer accept Bitcoin for vehicle payments. More than $8 billion in leveraged positions were liquidated across crypto futures markets within 24 hours, with over 800,000 trader accounts liquidated, Bitcoin futures alone accounting for roughly half of the total. The event was severe enough that the exchange most affected, Binance, briefly experienced its own outage, compounding the cascade for traders unable to manage positions during the most violent phase.
November 2022 — The FTX Collapse
FTX's native token FTT fell from $22 on November 7 to under $5 on November 8, an 80% one-day collapse, after reports surfaced that Alameda Research's balance sheet was substantially backed by FTT itself, triggering a bank run that saw 90% of all FTT withdrawn within days. Bitcoin fell roughly 10% and Ether roughly 15% on November 8 alone as the news broke, with the broader decline extending Bitcoin from approximately $20,500 to $15,600 and erasing roughly $236.7 billion from total crypto market capitalisation within two days. Unlike the other cascades in this article, the FTX event combined a pure leverage cascade with a genuine counterparty solvency crisis, since a meaningful share of the capital lost was customer deposits commingled and ultimately unrecoverable, not merely leveraged positions forcibly closed.
August 2024 — The Yen Carry-Trade Unwind
A sudden unwinding of the Japanese yen carry trade, a strategy in which investors borrowed cheaply in yen to fund purchases of higher-yielding assets globally, triggered a violent, cross-asset deleveraging event in early August 2024. Bitcoin fell sharply within a single week, one of its sharpest weekly declines of that year, as the same global liquidity withdrawal that hit Japanese and US equities simultaneously forced leveraged crypto positions to unwind. The event illustrated a mechanism distinct from the other three cascades in this article: the trigger originated entirely outside crypto, in a foreign exchange and global rates market dynamic, and crypto's leveraged derivatives markets transmitted and amplified a shock that began thousands of miles from any crypto-specific news.
October 10, 2025 — The Largest Cascade in History
Following a Trump administration announcement of forthcoming 100% tariffs on Chinese imports, crypto derivatives markets, entering the event with a record $217 billion in open interest, experienced the largest single-day deleveraging event in crypto history. Total liquidations reached $19.13 billion within 24 hours, affecting more than 1.6 million trader accounts, roughly nine times larger than any previous single-day liquidation total. Bitcoin fell approximately 14.3%, from $122,574 to roughly $105,000. Aggregate open interest across major exchanges fell 43% in the 24 hours following the crash, from $217 billion to $123 billion, with Hyperliquid alone seeing its open interest fall 57%, from $14 billion to $6 billion, as leveraged positions were forcibly unwound at a scale the venue had never previously processed. This event is modeled in full mechanical detail in DN's Algorithmic Pressure Gauge.
The Four Cascades, Compared
| Event | Date | Liquidated | Trigger | Trigger Origin |
|---|---|---|---|---|
| Black Wednesday | May 19, 2021 | $8B+ / 24hrs | China crypto crackdown + Tesla BTC reversal | Crypto-specific + corporate |
| FTX Collapse | Nov 2022 | $236.7B market cap / 2 days | Exchange insolvency, FTT bank run | Crypto-specific (counterparty) |
| Yen Carry Unwind | Aug 2024 | Sharp weekly decline | Global FX/rates deleveraging | External macro shock |
| Oct 10 Cascade | Oct 10, 2025 | $19.13B / 24hrs | 100% China tariff announcement | External geopolitical shock |
The pattern across all four: the trigger varies enormously, crypto-specific regulatory news, a corporate reversal, an exchange's own insolvency, a foreign exchange market thousands of miles away, but the mechanical transmission once a leverage cascade begins is identical every time. A price move breaches the maintenance margin of the highest-leverage tier, the exchange's automated engine force-sells regardless of impact, and that forced selling pushes price into the next tier's threshold. The trigger is unpredictable. The mechanism, once triggered, is not.
Model distributes aggregate open interest across $2,500 price bands using a leverage-tier decay curve calibrated against documented cascade proportions (Oct 10, 2025: $217B OI, 43% contraction in 24hrs). Illustrative model, not a live data feed. Cross-reference with Coinglass or exchange-native liquidation heatmaps for real-time figures before trading.
Illustrative relative OI shares across the five venues, recalculated against your slider input above. Hyperliquid's share reflects its rapid growth as a perpetual DEX since 2024; dYdX reflects its smaller, more legacy footprint in this comparison set as of mid-2026.
How to Actually Read a Liquidation Heatmap
Distance to the nearest cluster matters more than its total size
A $5 billion cluster sitting 25% away from current price is, practically, far less actionable than a $2 billion cluster sitting 3% away. The closer a cluster sits, the smaller the price move required to trigger it, and the more likely ordinary volatility, not a major news event, is enough to set off the cascade. This is the single most common mistake newer readers of liquidation heatmaps make: fixating on the largest dollar figure rather than the nearest one.
Clusters on both sides simultaneously signal the most violent potential moves
When large long and short clusters sit close to current price on both sides, simultaneously, the setup resembles a coiled spring: a move in either direction has a clear, immediate amplification mechanism waiting for it. This was visible ahead of several of the cascades described above, where open interest had built to record levels on both sides of price before the actual trigger arrived.
Exchange-level OI concentration changes which venue leads a cascade
Because liquidation engines operate independently per exchange, a cascade frequently begins on whichever venue has the largest concentration of leveraged positions closest to the trigger price, then spreads to other venues as the initial forced selling moves the broader market price that other exchanges reference for their own liquidation calculations. Hyperliquid's 57% open interest contraction during the October 2025 cascade, the largest percentage contraction of any major venue that day, illustrates how concentrated leverage on a single, rapidly-growing platform can produce a disproportionate share of a broader cascade's damage.
Positioning Around the Cascade, Not Inside It
The practical use of this information is not predicting exactly when the next cascade fires, which remains genuinely unpredictable given how varied the four historical triggers above actually were. It is sizing and placing your own positions with explicit awareness of where the mechanical pressure points sit. Avoid placing a stop-loss exactly at a major cluster's price level, since that is precisely where the heaviest forced-selling volume will already be concentrated, often producing a worse fill than a stop placed slightly beyond it. When a large cluster sits very close to current price, treat that as a signal to reduce leverage rather than increase it, regardless of which direction you believe the next move will go, since proximity to a cluster increases the odds that ordinary volatility, not your own thesis being wrong, is what closes your position.
Trade on platforms that make this data genuinely visible rather than obscured. Bybit and BloFin both provide transparent liquidation and open interest data alongside competitive funding rates, letting you check cluster proximity before opening a position rather than discovering it the way the heatmap above describes. OKX offers a comparably deep third option with strong derivatives liquidity. Open positions where your liquidation data is visible, not hidden.
What This Atlas Does Not Claim
Knowing where clusters sit does not predict whether or when price will reach them. All four historical cascades in this article were triggered by news events that were not predictable in advance, even though the mechanical cascade that followed each trigger was. This tool addresses the second half of that equation, not the first.
Exchange-reported open interest figures vary in methodology and timeliness across venues, and aggregating them into a single model necessarily involves estimation. Treat the relative proportions and price-band shape as directionally informative, not as precise, audited figures.
The Bottom Line: The Map Was Always Public. Almost Nobody Was Taught to Read It.
Four of the largest cascades in crypto history, spanning a corporate Bitcoin reversal, an exchange's collapse, a foreign currency market thousands of miles away, and a presidential tariff announcement, all transmitted through the exact same mechanical process once triggered: leveraged positions breaching maintenance margin, forced selling, and that selling pushing price into the next tier. The triggers were unpredictable. The transmission mechanism was not, and it remains visible in public data to anyone who knows to look for it.
See DN's Algorithmic Pressure Gauge for the full tier-by-tier compounding simulator calibrated against the October 2025 event, and the DN Market Maker Power Index for the specific firms whose positioning frequently sits on the other side of these clusters.
Frequently Asked Questions
A liquidation cluster is a price level where a large dollar value of leveraged positions would be forcibly closed if price reaches it, because that level corresponds to the maintenance margin threshold for a significant concentration of open positions at a given leverage tier. Clusters form because traders tend to use round-number leverage levels (5x, 10x, 25x, 50x, 100x) and round-number entry prices, causing liquidation thresholds to naturally bunch at predictable price distances from where positions were opened.
More than $8 billion in leveraged crypto positions were liquidated within 24 hours on May 19, 2021, across more than 800,000 trader accounts, as Bitcoin fell approximately 30% from roughly $43,000 to $30,000 within about 12 hours. The crash was triggered by Beijing reiterating its crackdown on crypto transactions, compounded by Tesla's recent reversal on accepting Bitcoin payments. Bitcoin futures accounted for roughly half of the total liquidated value.
The November 2022 FTX collapse combined a leverage cascade with a genuine counterparty solvency crisis. FTT fell from $22 to under $5 in a single day after revelations about Alameda Research's balance sheet triggered a bank run, with Bitcoin falling roughly 10% and Ether roughly 15% on the worst single day, and the broader market losing approximately $236.7 billion over two days. Unlike a pure leverage cascade, a meaningful share of the capital lost in the FTX collapse was customer deposits that had been commingled and were ultimately unrecoverable, not simply leveraged trading positions that could theoretically be rebuilt.
Following a Trump administration announcement of 100% tariffs on Chinese imports, crypto derivatives markets, entering with a record $217 billion in open interest, experienced $19.13 billion in liquidations within 24 hours, affecting 1.6 million trader accounts, roughly nine times larger than any previous single-day total. Bitcoin fell approximately 14.3%, from $122,574 to roughly $105,000. Aggregate open interest fell 43% in the following 24 hours, with Hyperliquid's open interest alone falling 57%, the largest percentage contraction of any major venue during the event.
Prioritize distance over size: a smaller cluster close to the current price is generally more actionable than a larger cluster far away, since it requires a smaller move to trigger. Watch for clusters on both sides of price simultaneously, which signals potential for a violent move in either direction. Avoid placing your own stop-loss exactly at a major cluster's price level, since that level will already have the heaviest concentration of forced-selling volume, often producing a worse fill than a stop placed slightly beyond it. When a large cluster sits very close to current price, treat it as a signal to reduce leverage rather than increase it.
The DN Liquidation Atlas models the distribution of aggregate open interest across five major derivatives venues (Binance, Bybit, OKX, Hyperliquid, dYdX) into price-band clusters around a given current Bitcoin price, then calculates a Cascade Risk Score combining two factors: how close the nearest large cluster sits to current price, and how concentrated the dollar value at that cluster is relative to typical bands. A higher score indicates either an unusually close cluster, an unusually large concentrated cluster, or both. The model is illustrative, calibrated against documented historical cascade proportions, and should be cross-referenced with live exchange or Coinglass data before making trading decisions.
Hyperliquid's open interest fell 57% during the October 10, 2025 cascade, from approximately $14 billion to $6 billion, the largest percentage contraction of any major venue that day, compared to a 43% aggregate contraction across all tracked exchanges. This likely reflects Hyperliquid's rapid growth as a leading perpetual futures venue heading into the event, which meant a larger concentration of recently-opened, potentially higher-leverage positions were present on the platform relative to more established venues with a longer history of position turnover and risk management adjustment.
A sudden unwinding of the Japanese yen carry trade, a strategy where investors borrowed cheaply in yen to fund purchases of higher-yielding assets globally, triggered a violent cross-asset deleveraging event in early August 2024. Bitcoin experienced one of its sharpest weekly declines of the year as the same global liquidity withdrawal that affected Japanese and US equities simultaneously forced leveraged crypto positions to unwind. This event is notable for originating entirely outside crypto markets, in a foreign exchange and global interest rate dynamic, illustrating that crypto's leveraged derivatives infrastructure can transmit and amplify shocks that begin far from any crypto-specific catalyst.
Embed grant: The DN Liquidation Atlas may be reproduced with attribution to decentralised.news.
DN-INTERNAL links to resolve: DN Algorithmic Pressure Gauge, DN Market Maker Power Index, DN Convergence Engine.
Sources: CoinGlass historical liquidation data, Webopedia "10 Biggest Liquidation Waves in Bitcoin's History," CoinDesk "Market Spotlight: Inside Crypto's $19 Billion Liquidation Event," Crypto.com market updates (Oct 2025), Yahoo Finance "9 Biggest Bitcoin Crashes in History" (Oct 2025), CoinPedia "Top 10 Biggest Bitcoin Crashes in History," Wikipedia "Cryptocurrency bubble" (FTX collapse data).
As of: June 2026. Not financial advice. Historical cascade data is verified; current heatmap figures are an illustrative model, not live exchange data.
About the Author
