Liquidation Cascades: Why Crypto Dumps So Fast (and Who Gets Wiped Out)

On Thursday, Nov 5, Bitcoin dropped about 6% and slipped under $76K. Not a crazy move by crypto standards. But according to CoinGlass, over $1 billion of leveraged positions got force-closed within 24 hours, and roughly 88% of that was longs. One snapshot caught $896M of it inside a single four-hour window.

This Sunday morning, the same CoinGlass dashboard shows $53M for the last 24 hours. Same market, about 24 times less carnage.

The difference isn’t some mystery whale. It’s leverage, plus what exchanges do when leverage goes wrong. Here’s how it works.

What actually happens when you get liquidated

When you open a leveraged position on a perpetual futures contract, you put up margin and the exchange lets you control a position several times bigger. At 10×, $1,000 controls $10,000 worth of BTC.

That margin is your buffer against losses. The exchange also sets a maintenance margin, a floor the buffer isn’t allowed to drop below. Binance spells it out: a liquidation triggers when your collateral (initial margin plus realized and unrealized PnL) falls below the maintenance margin.

At that point nobody asks you anything. The exchange closes the position for you. That’s a liquidation, or getting “rekt” if you spend too much time on crypto Twitter.

Two details people miss:

• It runs on mark price, not the last trade. Mark price blends prices from several spot exchanges with funding data, so a weird wick on one exchange is less likely to take you out. Less likely, not impossible.
• Bigger positions get less leverage. Exchanges raise the margin requirement as the position grows. The max leverage on the banner only applies to small positions.

How much room does leverage actually leave you?

This is the chart that should sit next to every leverage slider. It shows how far price has to fall before a long is liquidated, using a simplified model: isolated margin, a 0.5% maintenance margin, no fees.

The rough rule: your cushion is about 1 ÷ leverage, minus the maintenance margin. Everything at 20× and above didn’t survive an ordinary red day this week. Real liquidation prices sit a little closer than this, because fees and the exchange’s own margin tiers eat into the cushion.

Now put that next to this week. Bitcoin had climbed back above $80,000 before Thursday’s drop. A 20× long opened at $80K had a liquidation price around $76,380, and Bitcoin went under $76K. The liquidation engine doesn’t care how good your thesis was.

From one liquidation to a cascade

One liquidated trader doesn’t move the market. Thousands of them at roughly the same price do.

When the engine closes a long, it sells. Binance describes it as one large immediate-or-cancel order dumped into the order book. Market sells eat through the bids and price drops. If plenty of other longs have their liquidation price a bit lower, price walks straight into them, and they usually do: people pile in around the same levels with the same 10×, 20× or 25× settings, so their liquidation prices bunch up. Those positions get market-sold too. Price drops again. Next batch.

That’s a liquidation cascade. Basically dominoes, except every domino is a market order.

This is why crypto crashes look like a staircase falling through the floor instead of a smooth slope. It’s also why so much happens so fast. In one CoinGlass snapshot from Thursday evening, $896M of the $1.08B liquidated over 24 hours had happened in the latest four hours.

It works in reverse too. When price rips higher, shorts get liquidated, and closing a short means buying at market. That’s a short squeeze. Aug 19, 2026 is a good example: $2.99B liquidated, and $2.74B of it was shorts.

What happens when the engine can’t sell fast enough

In a really violent move the order book can thin out so quickly that the liquidation order doesn’t fill before the position’s losses exceed its margin. Exchanges have backup layers for that:

Step 5 is the wild one. With auto-deleveraging, if the insurance fund can’t take a bankrupt position, the exchange closes part of the winning positions on the other side, starting with the most profitable and most leveraged ones. So yes, you can get kicked out of a trade that was working. Binance says it tries hard to avoid ADL, and also admits it can’t avoid it completely given crypto volatility and the leverage on offer.

Don’t take the headline numbers literally

Every crash comes with a “$X billion liquidated” headline. Three things to keep in mind when you read one.

It’s position size, not money lost. “$1.26B liquidated” means positions with a combined notional value of $1.26B were force-closed. At 10× leverage, a $10,000 liquidation might be backed by $1,000 of actual margin. The losses are real, but they’re a fraction of the headline.

The number depends on when the screenshot was taken. Most dashboards use a rolling 24-hour window. The same selloff was reported as $1.08B on Thursday evening and $1.26B the next day. Neither is wrong. On CoinGlass’s all-time list of single calendar days, which starts at about $1.1B, Thursday doesn’t even make the cut.

The real totals are probably higher. CoinGlass notes that since April 24, 2021, Binance has published at most one liquidation order per second, and OKX has done the same since September 2021. When thousands of positions blow up in the same second, most of them never show up in the public feed.

Notice how much of that list is from 2021? Part of it is real, since 2021 was peak “100× on everything” energy. Part of it is that two of the biggest exchanges stopped reporting every liquidation that year, so newer events are probably undercounted. The record still belongs to Oct 11, 2025, which CoinGlass ties to the US tariff hike on China, at $19.16B. That’s about 15 times Thursday’s total.

What to watch (no predictions, promise)

Liquidation data doesn’t tell you where price goes next. It tells you how much fuel is lying around. A few things worth checking:

• Open interest. The total value of open futures positions. CoinGlass shows about $131B across exchanges this weekend. Rising open interest into a rally means leverage is stacking up. A sharp drop during a crash means leverage just got flushed out.
• Funding rates. When funding stays positive for a long time, longs are paying to stay long and the crowd is leaning one way.
• Liquidation heatmaps. These estimate where liquidation prices cluster. They’re models built on guesses about entry prices and leverage, not a list of real orders, so treat them as rough.

After a flush, a lot of the fuel is gone, which is one reason markets often calm down after a cascade, like this weekend’s $53M day. Often. Not always.

If you use leverage anyway

Liquidations are just math, applied all at once. Once you can see the math, the next “$1B wiped out” headline looks a lot less mysterious.