Data insight · Sep 23, 2026 · 8 min read
Two things happened at once this autumn. Traders who had bet on bitcoin falling were forced to buy it back, and money poured back into the US spot ETFs after most of a year of net outflows. The data shows which one lasted.
For most of 2026 the US spot bitcoin ETFs were shrinking. Net flows were negative in January and February, positive in March and April, then turned hard: $2.4 billion left in May and $4.5 billion in June, the worst month the funds have had. From the start of the year to mid-August, the net was an outflow of about $4.7 billion.
Futures traders read the same tape and leaned the same way. With bitcoin stuck between $63,000 and $65,000 through late July and early August and the Fear & Greed Index in the twenties, betting on further falls was the comfortable trade. That comfort is what the next five weeks punished.
A trader who sells bitcoin futures short profits if the price falls. Most do it with borrowed money, putting up a fraction of the position as margin. If the price rises far enough against them, the exchange closes the position for them before the margin runs out, which means buying the bitcoin back at the market.
That buying pushes the price up, which pushes the next group of shorts past their limit, which forces more buying. When a lot of traders are leaning the same way, one move up can set off a chain of forced purchases that has nothing to do with anyone wanting bitcoin.
A short that is forced out has to buy. Every forced buyer pushes the price into the next one.
That chain went off on 19 August. Futures exchanges force-closed $1.36 billion of bitcoin short positions that day, the largest day in the six months of liquidation data on our liquidations chart. Another $427 million followed on 20 August, when bitcoin rose almost 10%, and $735 million on 21 August. In three days, $2.5 billion of shorts were closed against $225 million of longs.
It happened again, smaller, on 21 September, when bitcoin cleared $85,000 for the first time since January: $558 million of bitcoin shorts closed that day in our series. Across every coin, CoinDesk counted $648 million of shorts liquidated in the 24 hours to that morning, out of $747 million in total.
From 17 August to 23 September, $4.85 billion of bitcoin shorts were liquidated, and $1.84 billion of longs. The forced buying was real, and it was concentrated in a handful of days.
A squeeze can start a rally. It cannot carry one for five weeks, because once the shorts are gone there is nobody left to force. What carried this one was the ETFs. They took in $3.5 billion in August, their best month of the year, and another $2.0 billion in the first fifteen trading days of September, according to Farside’s daily table. From 17 August to 22 September the net was $5.1 billion in.
The two biggest days came at the end. On 21 September the funds took in $999 million, the largest day of 2026 and one of the ten largest since they launched in January 2024. BlackRock’s IBIT took $381 million of it and ARK and 21Shares’ ARKB $289 million, the biggest day that fund has ever had. Another $715 million followed on 22 September.
Unlike a squeeze, an ETF inflow is a purchase of bitcoin that stays purchased. Our ETF holdings chart shows the funds went from 1,227,587 bitcoin on 14 August to 1,274,685 on 22 September: 47,098 more coins, bought in the market and held.
The funding rate on perpetual futures shows how hard traders are leaning long. When it is high, longs are paying shorts every eight hours to keep their bets open. In the early-2021 run it averaged 43% a year, and in the run-up to March 2024 it averaged 31%.
From 17 August to 20 September it averaged 4.1% a year on our funding chart. Bitcoin rose 37% on a market that was barely paying for leverage: the buying was coming from the funds and from shorts being forced out, not from traders borrowing to go long.
That changed at the very end. Funding jumped to 21% annualised on 21 September and was still 13% the next day, the first sign of leveraged longs chasing the move. Open interest across the major futures venues grew from $27.8 billion on 10 August to $38.4 billion on 22 September, with CME, where US institutions trade, up from $6.8 billion to $9.6 billion. Some of that CME growth is likely the basis trade that runs alongside ETF inflows: buy the fund, sell the future, collect the gap. Some of it is new leverage, and new leverage is what the next squeeze, in either direction, is made of.
Every day the network pays out about 450 new bitcoin in block rewards. Between 14 August and 22 September that came to roughly 17,500 coins. Over the same stretch the ETFs added 47,098. The funds alone bought about two and a half times everything miners produced.
That is the arithmetic behind a rising hashprice, which went from $30.50 per petahash per day on 14 August to $40.31 on 22 September. It is also the arithmetic that favours the miner who keeps the coins. When a buyer with that much appetite shows up, the coins already in the wallet are the scarce ones.
Squeezes end, and they can run the other way: the $802 million of longs force-closed on 2 June is on the same chart. The flows are the part worth watching, and they are on the Data section every day.
The funds bought 47,000 bitcoin in under six weeks. The whole network mined about 17,500.
Firsthand Bitcoin sells and hosts mining hardware, including this machine. No manufacturer, distributor or affiliate programme paid for or reviewed this page and we take no commission on the links above. Historical figures are computed from daily bitcoin price and network hashprice, each day valued at its own prices. Nothing here is investment advice.