- A Solana-aligned research group has put a figure of roughly $18bn of liquidations during the 10 October 2025 crash in front of the FCA and other regulators.
- Most exchanges publish a throttled feed that reports one liquidation per second per market, so published totals are known to understate the real number by an unknown margin.
- The group’s own conclusion, that transparent venues held up while opaque centralised ones failed, isn’t what the public records show.
A number has been travelling around since last October: $18bn of positions wiped out in fourteen hours, $3.21bn of it in a single minute. It’s now sitting in front of the Financial Conduct Authority. It is also, in a strict sense, not a number anyone can verify.
Regulators are being asked to draw lessons from the worst day the crypto derivatives market has had, and the evidence they’re working from is incomplete in a way that isn’t obvious from the outside. Whatever rules come out of that process will apply to UK holders, because the FCA is one of the bodies receiving the letter. It’s worth knowing how solid the foundations are before the building goes up.
There’s a second thing worth knowing. The largest, most-quoted figure in this debate was produced by a research group aligned with one particular , and it happens to support a conclusion favourable to that blockchain. That doesn’t make it wrong. It does mean it deserves the same scrutiny as any other interested party’s number.
What a liquidation actually is
Start with borrowing, because that’s all this is underneath.
A lot of crypto trading is done with leverage. You put down a deposit, called collateral or margin, and the exchange lets you control a position much larger than that deposit. Put down $1,000 and you might control $10,000 of bitcoin. If the price moves your way, you make ten times what you would have. If it moves against you, you lose ten times as fast.
The exchange isn’t doing this out of generosity. It’s lending you the difference, and it wants that loan back. So it sets a threshold: if your losses eat into your deposit past a certain point, it closes your position automatically and sells whatever you were holding to cover what you owe. That forced closure is a liquidation. You don’t get a phone call and you don’t get a choice.
Now picture that happening to hundreds of thousands of people in the same few minutes. Each forced sale pushes the price down slightly, which pushes the next tranche of positions past their threshold, which triggers more forced selling. That’s a liquidation cascade, and it’s what happened on 10 October 2025.
Why the published totals are wrong, and why everyone knows it
Here’s the part that almost never makes it into the coverage.
Exchanges publish liquidations through a public data feed. Most of the major ones throttle that feed: it reports a maximum of one liquidation per second, per market. Not one per second per trader. One per second, full stop, for that entire trading pair.
On a normal Tuesday that’s fine, because liquidations trickle in. In the middle of a cascade, when thousands are firing every second across every market at once, the feed reports one of them and silently drops the rest. The number that comes out the other end is not a count. It’s a sample, taken at a rate that has nothing to do with how many events actually occurred.
This is not a secret or a conspiracy. It’s a documented design choice, made to keep data feeds from overwhelming the systems that consume them. But it has a consequence that matters enormously here: every headline liquidation total from that day is a floor, not a figure. The real number is higher by an amount nobody outside the exchanges can calculate.
So when you see $18bn, or $19bn, or the various other totals that circulated, understand what you’re looking at. Those are reconstructions built on throttled data. They could be substantially understated. They cannot be checked.
What the public records do show
Where genuinely open data exists, the picture is more interesting than the slogan built on top of it.
On Hyperliquid, a decentralised derivatives venue, researchers were able to reconstruct a large auto-deleveraging event. Auto-deleveraging, or ADL, is the mechanism of last resort. When liquidations happen so fast that the losing side can’t cover what it owes and the venue’s own safety buffer runs dry, the exchange starts closing out profitable traders’ positions to balance the books. It’s different from an ordinary liquidation: that closes a losing position after the collateral runs out. ADL closes a winning one because someone else’s collateral ran out. If you were right about the market and still got closed, that’s ADL.
Aave, a lending , showed deficits and delays in its , the feeds that tell a protocol what an asset is currently worth. When those lag during a fast move, the protocol is making decisions on stale prices.
And the European Securities and Markets Authority later found that Binance’s internal collateral pricing amplified the forced selling on its own venue.
Read those three together and the honest conclusion is that transparency let us see the mechanics of the stress. It didn’t prevent it. Open venues broke in ways we can describe precisely. Closed venues broke in ways we can only infer. That’s a real difference, and it’s an argument for better disclosure. It is not an argument that one category of venue is safe.
The claim with an interest attached
The Solana Research Institute is a research group aligned with the Solana ecosystem. Its 14 August post revived a July open letter from Angus Scott to the FCA and other regulators, and the framing is that opaque centralised venues failed while transparent on-chain finance kept working.
Applying our own rule here: judge the effect, not the motive. We have no idea what anyone intended. The effect is that the largest circulating estimate of the damage, headed for regulators, was produced by a party with a stake in the conclusion it supports, using data everyone agrees is incomplete. Regulators read submissions like this all the time and are generally alert to who sent them. Readers seeing the $18bn figure recycled in a headline usually aren’t told.
None of this touches the people on the other end of those positions. Tens of thousands of traders were closed out inside a few hours, many of them on venues where the pricing mechanics were working against them in ways they had no way of seeing. What happened to them is the consequence of infrastructure decisions made well above their heads, and the useful question is what the firms and the regulators do about it.
What to watch
Whether any regulator requires exchanges to publish complete liquidation records rather than throttled samples. That single change would turn every future debate of this kind from competing estimates into a matter of fact, and it’s the concrete thing to look for in whatever the FCA and ESMA produce next.
And in the meantime, when you see a liquidation total quoted anywhere, two questions. Where did the data come from, and who assembled it. If the piece doesn’t say, that omission is usually doing some work.