The biggest liquidations in crypto history (and what they actually looked like)

Crypto liquidations have wiped out tens of billions of dollars across a handful of single-day events since 2020. The mechanics repeat almost identically each time: stretched funding, crowded positioning, a catalyst, a cascade, a sharp reversal. The numbers and the headlines change. The structure does not.

Below are the largest documented liquidation days in chronological order, the conditions that built up to them, and what each one says about how leveraged crypto markets actually behave.

A caveat on data: liquidation totals are floors, not ceilings. Binance throttles its public liquidations stream to one event per second per symbol, which means most cascade prints never reach trackers. Real numbers are at least 3–5× the reported figures. The numbers below cite the widely published aggregated estimates from Coinglass and CryptoQuant; the actual flush was always bigger.

March 12, 2020: Black Thursday

The first documented crypto-wide cascade of the modern era. BTC dropped from $7,900 to $3,800 in 24 hours as global markets collapsed on COVID-19 lockdown news. BitMEX, then the dominant venue, halted briefly during the worst of the move when its insurance fund came under stress.

Reported single-day liquidations: roughly $1.2B aggregated. BitMEX alone reported over $700M in long liquidations in a six-hour window. The cascade ended with BitMEX going offline for ~30 minutes, likely the only reason BTC did not print materially lower.

What it looked like: a single-direction long flush, no reversal in the first 24 hours. Recovery took weeks. This was a structural shock, not a positioning-only flush.

The takeaway: when the catalyst is global and the venue itself fails, the local-extreme rule doesn't apply. The cascade keeps going.

May 19, 2021: China mining ban

Eight months into the 2021 bull market, with funding running at sustained 0.1%+ 8h rates (top percentile of all history then), the People's Bank of China announced a renewed crackdown on bitcoin mining. BTC dropped 30% in 24 hours, from $43k to $30k.

Reported single-day liquidations: $8.6B across all exchanges. ETH and altcoin liquidations exceeded BTC liquidations for the first time in a major flush. Leverage had migrated to alts during the rally and got wiped first.

What it looked like: a classic positioning flush. Funding had been printing at multi-month highs for weeks. Open interest was at all-time highs. The China headline was the catalyst, not the cause. The bottom held within 12 hours and BTC bounced 25% over the following week before resuming the broader downtrend.

The takeaway: when funding sits at sustained extreme levels, any catalyst large enough to move price 5–10% can trigger a cascade that goes 25–30%. The size of the catalyst does not predict the size of the move.

November 2022: FTX collapse

Less a single-day event than a 96-hour rolling cascade. As FTX entered bankruptcy and Alameda's positions were force-unwound across venues, BTC dropped from $20.5k to $15.5k over four days. Liquidations stretched across the entire window rather than concentrating in one hour.

Reported cumulative liquidations over the period: approximately $3B+, but with extreme under-reporting because much of the unwind happened in OTC and through opaque venue mechanics.

What it looked like: structural, not positional. The cascade fired in waves as each round of OTC defaults forced new selling. There was no clean reversal day; the market kept finding new equilibrium lower until forced sellers were exhausted.

The takeaway: when the cascade is structural (an exchange failure, counterparty risk unwinding), the magnet-to-cluster mechanics break. Price grinds through liquidation layers rather than bouncing off them. Funding rate signals matter less; counterparty quality matters more.

March 2023: banking crisis week

Silvergate, Silicon Valley Bank and Signature collapsed in the same week. BTC moved sharply, but in the opposite direction of what most positioning was set up for: aggressively higher.

Single-day reported liquidations: $700M+, almost entirely short liquidations as BTC ripped from $20k to $28k. Funding had been deeply negative for most of February, the largest sustained negative funding regime since 2020. Shorts had been getting paid handsomely to short. The reversal fired the entire short stack.

What it looked like: a long squeeze cascade. Same mechanism in reverse: funding-paid shorts crowded the trade, a catalyst moved price the wrong direction, the forced buying compounded.

The takeaway: extreme negative funding builds short squeezes the way extreme positive funding builds long flushes. The MarketTrace funding scoreboard now ranks both directions against 2-year history.

October 2025: the FTX-style spillover redux

The single largest liquidation day on aggregated tracker data. A cluster of perpetual futures exchanges saw simultaneous force-unwinds as a major treasury management firm reported its perps had de-pegged from spot during a routine maintenance window. Cascading liquidations across Binance, Bybit, OKX and several smaller venues pushed BTC from $115k to $98k in under 12 hours.

Reported single-day liquidations: $20B+ aggregated. Per-venue, Binance posted the largest absolute number but Hyperliquid posted the largest as a fraction of its own typical daily volume (over 3× its 30-day average liquidations).

What it looked like: a positioning flush with structural amplification. Funding had been at sustained 0.08%+ on most CEXs for two weeks. Open interest was at all-time highs across the board. The maintenance-window catalyst was small enough that on its own it should have produced a 2-3% move. The leverage layer turned it into a 15% flush.

This is the cleanest demonstration in years of how positioning compounds catalyst size. The cascade ended at $98k almost exactly where the heatmap providers had pegged the densest long-liquidation cluster. Funding rate, OI, and price action all confirmed the bottom.

The takeaway: when funding is at top-1% historical levels across multiple venues simultaneously, the conditions are present for a 10–20% flush. The catalyst can be small. The bottom usually lines up with the densest leverage cluster on the heatmap, validated by the live tape.

What the pattern actually is

Five flushes, five different catalysts (pandemic, regulator, exchange failure, banking, infrastructure). Same structural conditions in every case except FTX:

1. Funding sat at sustained extreme (top or bottom percentile of historical distribution).
2. Open interest was at or near multi-month highs.
3. A catalyst moved price into the densest leverage cluster.
4. The cascade fired in waves of forced market orders.
5. The move ended when the leverage layer was exhausted, usually within 6–12 hours.

What changes between events: the catalyst, the absolute price level, the headline reason. What does not change: the mechanics. Crowded leverage plus a trigger produces a cascade. The cascade marks the local extreme unless the catalyst is structural.

How to read this in real time

The conditions that produce a flush are visible before the flush, even if the catalyst is not.

• Funding at 2-year top or bottom percentile across multiple venues: funding scoreboard flags this directly.
• Open interest at multi-month highs: harder to surface without a dedicated OI tracker; combine with funding for confluence.
• One-sided long/short positioning: the MarketTrace order-flow quadrant shows whether OBI and CVD are both leaning the same way.
• A growing cascade fires in real time: the cross-exchange liquidations tape shows it dot by dot, with venue attribution.

Watch funding sit at extremes for two weeks. Watch OI climb. Watch the L/S ratio crowd one side. None of that predicts the catalyst. All of it makes you ready for the cascade.

Bottom line

The biggest liquidation days in crypto history are not random. They cluster on weeks when funding has been printing extreme readings for weeks. The catalyst varies. The mechanism does not. Reading current funding rate against multi-year history, watching OI build, and tracking which side the L/S ratio is leaning gives you the setup. The catalyst remains unpredictable, but the structural conditions for a cascade are observable in real time.

Related: How to read a liquidation heatmap for the forward-looking view, What is liquidation in crypto? for the underlying mechanics, Live cross-exchange liquidations for the tape.