以太坊空头清算潮：2.7亿美元爆仓背后的机制与误判

On the 20th of August at 4:51 in the morning, a trader on Hyperlquid known as PensionUSDT. 50,000 ETH. 12 seconds later, the position no longer existed. About $24 million gone, which is about half of everything that trader had ever made shorting crypto. liquidated in four tranches with the final 1,417 ETH absorbed by the exchanges backs stop fund and they were one of more than 172,000 traders wiped out in the same window.

Over $2.7 billion of shorts liquidated inside of 24 hours. More than 90% of everything closed was somebody betting lower and Ethereum accounted for around $1.13 billion of it. the largest one-sided short liquidation event since November of 2021, which means the trade that had been consensus for the better part of a year turned around with a vengeance. So, today we're breaking down how Ethereum is turning things around, why the bearish thesis behind all that short positioning was focused on the wrong thing entirely, and what the SEC proposed one day before the market decided to decimate ETH shorts.

My name is Lewis and this is the Coin Bureau. Now, let's start with what actually happened with ETH during that short liquidation cascade between the 19th and the 21st of August. The headline figure was somewhere close to $3 billion of liquidations in a single 24-hour window with 48 hour commulative totals running as high as $4 billion. Longs accounted for just 300 or so million of that. And if you got liquidated on a long while prices pumped 20 plus%, honestly, I'm not sure what to tell you.

But anyway, everything else was shorts. And this wasn't a single spike either because a second 24-hour window into Friday the 21st added another $1.24 billion of which around 1 billion was once again short positions being forced closed. Now, these epic liquidation events are a core feature of crypto, but there's some important context here. Most notably, in contrast to what you might see or hear elsewhere, this was not the biggest liquidation event in crypto history.

The October 2025 crash was far larger in gross dollar terms at roughly 19.2 billion, but that was a mixed event with longs and shorts both getting torched. What made August different was the concentration. Nine out of every $10 liquidated belonged to somebody positioned against the market. And on that specific measure, you have to go back to November of 2021 to find anything comparable. So then why did Ethereum shorts bear the brunt of it?

Well, the technical answer is positioning plus depth. And both were pointing the same way. Ethereum carried heavily netsh short open interest relative to Bitcoin heading into the week with perpetual funding rates sitting near neutral to negative across the major venues for weeks on end. Bitcoin's funding by contrast was closer to flat. Meanwhile, Bitcoin had spent 6 weeks between the 8th of July and the 18th of August grinding sideways in a tight range, which is exactly the kind of compression that encourages systematic traders to keep fading the top of it.

But the real problem was the supply side of the order book because ETH held on exchanges fell roughly 15% from early June to mid August, dropping from around 7.7 million ETH down to about 6.54 million. Less spot ETH sitting on exchanges means less resting supply available to absorb a wave of short covering and market makers were apparently positioned net short themselves above the $19,50 to $2,000 area. So when the buying arrived, there was nothing but a vacuum up there.

A thin book, crowded short positioning, and One Direction. There's also a funny bit of behavioral detail buried in the onchain data here. Wallets holding 1,000 ETH or more had trimmed their holdings by roughly 1.7 million ETH since the 20th of May. And that distribution fostered a consensus among bears that supply overhang would cap any rally. What they missed though was that wallets holding between 1 and 10 ETH had grown their share of supply to 4.52% over the same stretch.

So the coins were truly just changing hands as opposed to being dumped on the market. And this is where we could go back to our short whale for a second because that trade tells us a lot about how confident this side of the market had become. That short was built over around 2 months at 3x leverage with an average entry near $1,700 and a notional exposure of around $16 million. The trader had a 23 win streak behind them and roughly $49 million in lifetime profits from shorting, including about $6 million on a $60,000 ETH short in June alone.

So, this is somebody who had been right repeatedly for a very long time. And just like that, the market humbled them in a big way. Now, the thing is, they weren't alone in the graveyard. One wallet had 1,800 Bitcoin of short exposure wiped for somewhere between 96 and 117 million. And the account known as Bowen 1476, lost about 71.5 million on 500 BTC, and a third handle went down for roughly 61 million. Ouch. What if you could trade real US stocks like Apple, Nvidia, or Tesla without leaving your crypto account?

Well, that's the idea behind our tokens from BitGet. These are tokenized stocks backed one to one by real shares. But the key difference is they are actually usable. You can trade them, use them as margin, and even earn dividends instead of just letting them sit in your wallet. So, if you want to check them out for yourself, sign up for BitGet using the link in the description or by scanning this QR code. But whatever the case, the ETH move was the violent one, running around 20% intraday against Bitcoin's 8%.

Now, market mechanics is one thing, but you only get positioning that lopsided when a bearish story has become the default assumption. And to be fair, the story was reasonably convincing. The argument against ETH was largely that layer 2 networks or L2s, the roll-ups that sit on top of Ethereum and process transactions more cheaply, have cannibalized mainet activity. Less activity on the base layer means fewer fees. Fewer fees means less ETH burned, and less ETH burned means Ethereum has no economic engine.

That thesis is what the ETH BTC ratio was pricing when it bottomed in late June, printing its weakest levels in the better part of a year. And again, to be fair, there's some substance here. Base layer fee generation has been running at something in the region of $700,000 a day with total L1 revenue nearer $330,000 for settlement network securing hundreds of billions of dollars. There's no sugar coating it. That is a small number, but this is the part where positioning got wrong.

Falling base layer fees weren't the symptoms of Ethereum losing. They were the deliberate output of a road map that has been executing for 2 years. It's a bit like a motorway operator who spends a decade widening the road and then scraps most of the tolls. Read the toll receipts on their own and the business looks like it's dying. Look at the traffic and at what's actually being carried and you're looking at national infrastructure.

EIP4844 made it cheaper for Ethereum's L2s to post their data back to the main network without clogging up normal transactions. Then in December 2025, the Fusaka upgrade introduced pure DOS, allowing Ethereum to handle much more of that L2 data at once. The result is that transaction fees on the major roll-ups now run consistently below 2 cents, a reduction of more than 90% with some measures of median L2 fees down over 95% against previous years.

And there's more coming, which was arguably very underpriced up until recently at least. Glamsterdam combines two major Ethereum upgrades, Glowaz and Amsterdam, and core developers are saying that it's the biggest overhaul of the network since the merge. The goal is to massively increase how much activity Ethereum can handle. It would raise the minimum gas limit per block to 200 million, more than triple today's level, while also changing how transactions are ordered and blocks are built so that more of the core processes happen directly inside