Amberdata's reconstruction of that October cascade makes the mechanics uncomfortable to look at. In the single worst minute, $3.21 billion was liquidated, and 93.5% of it was forced selling. Across the month, longs made up 83.9% of all liquidations — a 5.2-to-1 ratio against shorts.
That isn't the market forming a view. It's a chain reaction, where each liquidated position becomes a market sell order that drags the next account across its own line.
October wasn't an outlier either. It was a magnification. Across 2025 as a whole, something like $150 billion in positions were force-closed, and longs ran ahead of shorts all year — on February 3 alone, $1.88 billion of long liquidations came through, about 65% of that day's total.
The pattern repeats because the habit behind it repeats.
Nobody thinks they're the overleveraged one
Divide the day out. Nineteen billion dollars across 1.6 million accounts is roughly $12,000 each. These weren't whales miscalculating. They were ordinary retail traders, sized as though the next 14% would arrive slowly and give notice.
The mistake I see most often isn't using leverage. It's treating leverage as a return multiplier and never once converting it into a distance.
People pick 10x because it sounds like a more serious version of 3x. Almost nobody then opens a chart and asks how often this asset has moved 9% in a day. That check takes about ninety seconds, and it would end most of these positions before they were ever opened.
The second mistake is trusting the stop to save you. During that cascade, bid-ask spreads blew out to more than a thousand times their normal width. A stop is an instruction to trade with somebody, and for a few minutes there wasn't anybody. Liquidation engines don't wait around for a good fill. They take what's there.
Size backwards
Before you set leverage, name two prices: the one that proves your idea wrong, and the one that's just this asset breathing. Then pick leverage so that liquidation sits well outside both. If those two prices are the same number, you don't have a trade. You have a coin flip with a fee attached.
Do it with a calculator open, not from feel. Take your entry, take your maintenance margin, and write down the actual number the exchange will act on. Then go look at how many times in the last two years the asset travelled that far in a single session.
My own floor on bitcoin is that liquidation should sit at least 25 to 30% away, which caps me somewhere near 3x. It's a boring number. Boring is what lets me still be in the position when the recovery arrives — and nearly every one of those 1.6 million accounts was eventually right about direction. They just weren't there for it.
Here's the part I think matters most. Leverage gets chosen in the moment, usually while a green candle is still forming, and that is the worst possible condition under which to do arithmetic. That's most of the reason I stopped choosing it by hand.
