How is a liquidation cascade different from ordinary liquidation? Isn't liquidation supposed to be a protocol protection mechanism?
You're right — liquidation is a protection mechanism. The problem is when the 'volume' of liquidation exceeds what the market can absorb, the protection mechanism becomes a damage amplifier instead.
Ordinary liquidation (normal market): ETH steadily declines 5-10% per day, liquidating a batch of positions. These positions are unhurriedly bid on by liquidators, ETH enters the market and is gradually absorbed, with limited overall market impact. The protocol is safe, and your position (if collateral ratio is adequate) is unaffected.
Liquidation cascade (extreme market): ETH drops 40% in 3 hours, thousands of positions simultaneously hit liquidation thresholds, massive volumes of ETH flood the market at once. Liquidators can't absorb everything even at full capacity — this either drives ETH even lower (triggering more liquidations), or insufficient liquidators cause the protocol to accumulate bad debt.
Essential difference: Single liquidation is 'surgery'; liquidation cascade is 'organ rejection' — the same mechanism produces opposite effects at different scales.
This is why MakerDAO after 2020's Black Thursday heavily strengthened liquidation incentives, introduced Liquidation 2.0 (Dutch auction), and required higher liquidation buffers — the historical lesson isn't that the liquidation mechanism is broken, but that when market crash speed exceeds the mechanism's design limits, more safety cushion is needed.
What exactly happened on 2020's Black Thursday? Why didn't MakerDAO prevent this problem in advance?
On March 12, 2020, crypto markets crashed sharply amid global financial market panic triggered by COVID-19. ETH fell from approximately $196 to $88 within 24 hours, a decline exceeding 55%.
What specifically happened: ETH rapidly fell through the liquidation lines of large numbers of MakerDAO CDPs (Collateralized Debt Positions). The protocol started triggering Liquidation Auctions — which should have worked as: liquidators bid to buy ETH, more bidders push prices higher, and Maker protocol recovers more funds.
Where the problem lay: Ethereum's network suffered severe congestion amid panic selling, with Gas fees surging to 10-20x normal levels. Many liquidators' (arbitrage bots') bidding transactions couldn't get on-chain, or by the time they did the liquidation had expired. Result: some liquidation auctions settled at '$0 bids' — meaning someone obtained ETH that should have been worth hundreds of dollars at near-zero prices.
Final consequences: MakerDAO accumulated approximately $3.7M in Surplus Deficit (protocol owed funds). MakerDAO's community voted in emergency to issue new MKR tokens, auctioned them on the market, and used the proceeds to fill the gap.
Why it wasn't prevented: At the time, MakerDAO's liquidation system (Liquidation 1.0) used English auctions (highest bidder wins), assuming liquidators would actively bid. Under extreme network congestion, this assumption failed. Afterward, MakerDAO upgraded to Liquidation 2.0 (Dutch auction — price starts high and descends, first acceptor wins), substantially improving the liquidation system's resilience.
If I'm a liquidator, is a liquidation cascade an opportunity or a risk for me?
Both — and the risk is often underestimated.
Opportunity side: During a liquidation cascade, large amounts of collateral enter the market at discounted prices. If you have sufficient liquidity and can front-bid during network congestion, you can earn enormous arbitrage returns in a very short time — this is why some liquidation bot developers publicly reported massive profits after Black Thursday.
Risk side:
Biggest risk: During liquidation cascades, those who earn the most are typically professional institutions with pre-built infrastructure (low-latency nodes, high Gas priority transactions, automated liquidation strategies). Individual investors participating ad-hoc in liquidation competition are unlikely to outcompete these institutions, and may end up buying ETH that continues to fall during the market crash.
If the market starts showing signs of a liquidation cascade, what can I do immediately?
Emergency action checklist in priority order:
Execute immediately (first 15 minutes):
Within your available action window (30-60 minutes): 4. If you have idle USDC or ETH, inject collateral directly into your MakerDAO Vault to raise the ratio 5. Or repay some USDS to the protocol (reducing liabilities), which also raises collateral ratio — faster if you have excess USDS on hand 6. If your position is highly leveraged and you can't add collateral, consider proactively closing part of your position (though there's a loss, it's better than forced liquidation's typical 5-15% discount penalty)
After the chaos subsides: 7. Analyze this risk event: was your collateral ratio buffer sufficient? Did automated protection trigger in time? Systematically raise future safety thresholds.
Core principle of prevention over response: In 99% of cases, setting adequate buffer (200%+) and automated protection during normal markets is far more effective than scrambling during a market crash.
March 2020 Black Thursday: The Most Severe Liquidation Cascade in DeFi History
On March 12, 2020, global financial markets were violently shaken by COVID-19 pandemic panic, and crypto markets were caught in the fallout. This day became the most important stress test in DeFi history.
Timeline:
Community response:
What this event taught DeFi: Liquidation system design must account for extreme network congestion scenarios; liquidation incentives must be high enough to cover extreme Gas fee environments; Surplus Buffer isn't optional but a necessary safety cushion. These improvements allowed MakerDAO to safely weather multiple subsequent market crashes (including the 2022 Luna collapse and FTX collapse) without similar bad debt events.
Liquidation Efficiency vs Liquidation Robustness: The Fundamental Protocol Design Trade-Off
Liquidation system design faces a fundamental trade-off: faster, more aggressive liquidation versus more robust but slower liquidation?
Aggressive liquidation (fast): Set lower liquidation lines (e.g. 110%), liquidating large positions immediately when ratio is hit. Advantage: very low bad debt risk (strong protocol protection); disadvantage: users' collateral easily liquidated during brief market fluctuations, poor user experience, potentially discouraging many reasonable borrowing activities.
Conservative liquidation (robust): Set higher liquidation lines (e.g. 145-150%) and only liquidate in batches, giving markets time to absorb. Advantage: user positions less likely to be liquidated during normal volatility; disadvantage: during extreme crashes, the protocol may already be in bad debt before liquidation can complete.
Liquidation 2.0 (Dutch auction) compromise: MakerDAO's Dutch auction starts from high discounts and gradually reduces them until someone accepts. This lets liquidators flexibly choose when to bid based on their costs — more resilient under network congestion than English auctions (bidding up from low price). Even when Gas fees are high, the Dutch auction structure can still attract liquidators to act when economics favor them.
Meaning for you: There's no perfect liquidation design, only trade-offs with different advantages under different market stress conditions. The best defense you can mount is to always keep your collateral ratio far above protocol requirements, ensuring that whatever challenges the protocol's liquidation mechanism faces, your position doesn't need to rely on it.