On the evening of October 10, 2025, Donald Trump announced 100 percent tariffs on all Chinese imports. Equities and commodities sold off — and then closed for the weekend. Crypto never closes.

Over the hours that followed, exchanges force-sold positions worth roughly $19 billion. It is the largest liquidation day this market has seen. Four days earlier, Bitcoin had set its all-time high.

Did that night make the bear market? No. It was the occasion. Trump's announcement set the date and nothing else. The cause of the crash was the leverage in the market: that decided how far it fell that evening. And the ground it fell on had been prepared much earlier. Bitcoin had peaked four days before, exactly where it has peaked after every halving. The market was running out of buyers.

One number sits awkwardly in that story. Bitcoin lost 7.2 percent that day. The worst day of the whole decline came four months later and cost nearly 14 percent — no tariffs, no cascade, no headline.

Leverage turned a sell-off into a cascade

Buy on credit and you post only a fraction of the purchase value as collateral. If the price falls far enough, the exchange sells on its own. It asks nobody, and nobody needs to be at a screen.

Every one of those forced sales pushes the price down. At the next account the collateral then falls short, and the exchange sells again. That is how the night ran, thousands of times over.

The $19 billion is not the money that was lost. It is the value of the positions that were closed. Post a tenth of the purchase value, get forced out, and you lose that tenth — the statistic records the full amount. CryptoQuant puts the actual losses on Bitcoin and Ethereum at roughly $2.3 billion.

The figure is uncertain in the other direction too. Major exchanges cap how many liquidations they report publicly, so they under-report precisely when the most are clearing at once.

Leverage does not only change the outcome in a crash. It also changes, in a calm market, what is left of a strategy.

On Binance the collateral was suddenly worth 65 cents

The stablecoin USDe is supposed to be worth a dollar at all times. On Binance it fell to 65 cents. Only there — on other venues it stayed well above.

Anyone who had posted it as collateral lost a third of their security while the same coin was worth almost a dollar everywhere else. Those people were not liquidated because they were wrong. They were liquidated because their collateral was mispriced at one venue for a few minutes. Attackers have exploited the same mechanism in decentralised markets for years: move the price source and you move other people's positions.

Binance published the sequence afterwards with its own timestamps. Few exchanges do. And Binance disputes that these price deviations set off the crash: the broad market, it says, had already bottomed before they began.

That is the account of the house in question. It does not change the fact that Binance then paid $283 million to the affected holders. Paying is conceding that damage occurred.

The small coins fell, Bitcoin held up

In that single day, the balance of the market shifted sharply towards Bitcoin.

The altcoin share of the total market shrank by more than a tenth in one day. On some venues, prices appeared that existed nowhere else: CoinShares documents a spike down to almost zero for ATOM on Binance. Anyone holding a tight stop there was filled at a price the market never had.

Three days later everything stood where it had before

This part is missing from the retrospectives.

The Arena Pulse condenses ten market indicators into one daily reading between 0 and 100. On October 10 it stood at 68 — the same range it had held all week. The next day it fell to 58. On October 13 it was back at 64.

Three days after the largest liquidation day in this market's history, the daily reading said what it had said before. Our traffic light for altcoin rotation also stayed green on the day of the crash and for two days after.

That is not a flaw in the measurement but a property of daily data. A reading built from closing prices cannot report an event that begins at a quarter past nine in the evening. And it averages away whatever has been bought back by the next morning. Five weeks passed before the Pulse first dropped into risk-off territory.

The speculators turned late as well. Hold a leveraged long and you pay the funding rate to the other side for as long as buyers dominate. It stayed positive through October 11 — so the long side was still paying. Only on October 12 did it turn negative. (Back then it drew on two venues, Binance and Bybit.)

The big money took longest of all:

The peak sits on the evening before the cascade, and it has not been reached since. Institutional money was still arriving during the crash. The sustained outflows only started afterwards.

In euros the all-time high was barely a high

The chart everyone saw is drawn in dollars. And the dollar was moving on its own: against the euro it lost 15.0 percent between February 2025 and January 2026. Price Bitcoin in euros and you are looking at a different market.

Both rows come from the same venue, so the comparison measures the currency rather than the data source.

In dollars the October high sat 17.4 percent above the January high. In euros it was 4.2. Compare daily highs instead of closes and 1.3 percent is left in euros, against 15.2 in dollars.

October was still the high in euros — but only just. And that explains why October 6 did not look like a peak: after nine months, a European holder stood barely above his old record.

The market was not quiet, though. Between January and October, euro prices spanned 52.9 percent; in April the price fell to 69,489 euros and climbed all the way back. There was plenty of movement. What was small was what remained at the end — and part of that came from the exchange rate rather than from Bitcoin.

For the fall that followed, the currency makes no difference. Since the October high the loss is 48.6 percent in dollars and 48.0 in euros. The exchange rate explains how high the high was, not the fall.

The high arrived at the usual point after the halving

Every four years the amount of new Bitcoin per block halves. Every time so far it ran the same way: about a year and a half up, a high, a deep fall.

That proves nothing. A date that fits a pattern does not explain why the selling happened on that particular evening — Trump was needed for that. It does explain why the market never climbed back: the buyers who carry a high like this had, in every previous cycle, been used up by that point.

The pattern also fulfils itself in part. When many holders expect a high in the fourth year, they sell into it and buy above the old high more hesitantly. An external shock then meets that thinner demand, and the expected break duly arrives. So the cycle is no law of nature — but no superstition either. It is a widely held expectation, and expectations move prices.

The worst day came four months later

October 10 cost 7.2 percent and got the headlines. February 5, 2026 cost nearly 14 percent and passed almost unnoticed. In the whole year after the high, it is the only day that fell harder than the record cascade.

A recovery came before it. By mid-January, Bitcoin stood well above the November low, and anyone watching only the price saw things settling down. But since October the market had been missing a large part of its credit and a large part of the bids in its order book. A market like that falls harder, even without fresh news.

What would prove this wrong

If Bitcoin climbs back above the high of October 6, 2025 without another halving in between, October was a deep correction rather than a cycle top. The prepared ground drops out, and what remains is a macro shock with a long tail.

The second point is closer. If someone finds a market measure that warned on a daily basis before October 10, the finding about the delay is wrong. So far the sequence argues against it: rotation out of altcoins into Bitcoin jumped on the same day. The Pulse, the traffic light and the funding rate all jumped later. None of them jumped earlier.

Occasion, cause, prepared ground

Occasion: Trump's tariff announcement. It set the date and nothing else.

Cause of the crash: the leverage in the market and the empty order books. They turned a sell-off into a cascade — and turned the market price, for a few minutes, into a question of which venue you happened to be on.

Prepared ground: a cycle running out of buyers. That is why the break held instead of reversing.

October 10 can be timed to the minute. That makes it the better story — not the whole explanation.

Frequently asked questions

What triggered the Bitcoin crash on October 10, 2025? Trump announced 100 percent tariffs on Chinese imports on a Friday evening. Equities and commodities closed shortly afterwards while crypto kept trading — and met the shock holding more positions on credit than ever before.

How much was liquidated on October 10, 2025? Roughly $19 billion, measured by the value of the positions closed. The money actually lost was far lower: CryptoQuant puts it at around $2.3 billion for Bitcoin and Ethereum.

Was October 10, 2025 the cycle top? No. The all-time high came four days earlier, on October 6, 2025. Market structure broke on October 10; the high had already happened.

Why did altcoins fall harder than Bitcoin? They trade thinner. When market makers withdraw, small coins lose their buyers first, and forced selling meets an empty order book. The altcoin share of the total market shrank by more than a tenth in that single day.

Was the October 2025 all-time high also an all-time high in euros? Yes, but narrowly. In dollars it sat 17.4 percent above the January high of the same year; in euros only 4.2 percent, and on daily highs just 1.3. The rest of the visible gap was the dollar's decline.

Did the four-year cycle cause the bear market? It is the prepared ground, not the occasion. The high came 534 days after the halving, between the 526 and 548 days of the two previous cycles. That explains why no recovery followed — not why the selling happened that evening.

---

This text describes market mechanisms and the data behind them. It is not investment advice, not a recommendation, and not a forecast. Historical patterns are no promise about the future.

Sources: Binance, daily candles for BTCUSDT and BTCEUR, retrieved August 17, 2026 (currency comparison) · Own daily series for price, market shares and the rotation traffic light (data provided by CoinGecko), retrieved August 17, 2026 · Arena Pulse, daily readings, retrieved August 17, 2026 · Own funding series (Binance, Bybit), retrieved August 17, 2026 · ETF flows: SoSoValue, as of August 14, 2026 · Reuters: Bitcoin down 5.5%, October 10, 2025 · Reuters: After record crypto crash, a rush to hedge against another freefall, October 13, 2025 · CoinDesk Data: How CCIX performed during crypto's largest liquidation event · CryptoQuant, analysis of realised liquidation losses, October 17, 2025 · CoinShares: Billions in liquidations — what happened? · Binance, statement on the market event of October 10, 2025 · Binance, announcement on compensation for USDe, BNSOL and WBETH · Halving dates: Bitcoin blockchain (block 840,000, April 20, 2024).

As of August 17, 2026.

Study the Past — Improve your Future 🥋