Data insight · Sep 29, 2026 · 6 min read
Funding went from 21% to negative in four days. Bitcoin held $84,000.
On 21 September, annualised funding on the bitcoin perpetual hit 20.7%, its highest reading of the whole rally, as traders borrowed to chase the price rather than being forced to cover a losing short. Four days later it was negative. Open interest fell with it. The price barely did.
Leverage built for two days, then it flipped
Bitcoin's squeeze in the third week of September ran on short covering and ETF buying, not on new leverage. From 17 August to 20 September, annualised funding on the bitcoin perpetual averaged 4.1%, as we covered in the squeeze and the flood. That changed on 21 September, when funding jumped to 20.7% annualised, the highest reading of the entire run, and held elevated at 13.0% on 22 September and 10.1% on 23 September, on our funding chart. For the first time since the squeeze began, traders were borrowing to chase the price rather than being forced to close a losing bet.
It showed up in open interest too: every open long matched by an open short, across the major futures venues, rose from $35.8 billion on 21 September to $38.6 billion on 23 September. Some of that was new leverage. Some of it was the basis trade that tends to grow alongside ETF inflows — buy the fund, sell the future, collect the gap.
The liquidations changed sides
The reversal showed up first in forced closures. On 21 September, exchanges closed $558.3 million of bitcoin short positions against $57.4 million of longs in our liquidations series — the squeeze itself, and the day funding turned. Two days later, on 23 September, it was longs being forced out: $131.1 million closed against $33.4 million of shorts, as bitcoin fell from an intraday high of $87,392 back under $84,000. Cointelegraph reported $280 million of long positions liquidated across the wider crypto market in four hours that afternoon, with the analyst Rekt Capital telling followers bitcoin needed to "stay above or at minimum successfully retest ~$82k" to avoid falling back into the $60,000–$80,000 range it had spent the summer in.
Another $82.8 million of longs closed on 24 September, $45.6 million on 25 September. By 26 September the futures market had gone almost quiet: $3.8 million of longs and $4.9 million of shorts, a fraction of the $615.7 million that had changed hands five days earlier. Bitcoin itself never came close to giving back the move. Its close never fell below $84,000 in the whole week from 22 to 28 September, even as the leverage behind it emptied out underneath it.
On 21 September, $558 million of shorts were forced out. Two days later, $131 million of longs were.
Funding went negative, and stayed near zero
Funding followed the liquidations down. It fell to 3.9% annualised on 24 September, then to −1.6% on 25 September — for one day, traders holding short were being paid, not the other way round. It has barely moved since: 0.02% on 26 September, 0.03% on 27 September, 0.004% on 28 September. Four days after its 21 September peak, the reading that had signalled leveraged longs chasing the rally was back to essentially nothing.
The total book of open contracts shrank with it. Across every major venue, open interest fell for four straight trading days after its 23 September peak, to $35.45 billion on 27 September, a drop of about 8%. crypto.news reported that Binance's own open interest alone fell roughly $500 million between 21 and 23 September, while its funding rate flattened to close to zero — the same retreat our own numbers show across every venue combined.
One line didn't follow the rest down
Open interest on CME, the venue where the open contracts are mostly US institutions and the ETF issuers' own hedges, kept climbing through the same four days that the total was falling: from $8.96 billion on 21 September to $10.03 billion on 27 September. It is a smaller book than the total, and nowhere near its own December 2024 peak of $22.7 billion, so this is not a record. But as the faster-moving, mostly offshore leverage came out of the market, CME's share of every open bitcoin future rose from about 25% on 21 September to 28% on 27 September, on our open interest chart.
That is a different kind of position than the leverage that just unwound. A trader on a perpetual future pays funding every eight hours and can be closed out by the exchange within minutes. A basis position built around an ETF holding, or an institution's own hedge, does not answer to the same clock.
The offshore book got smaller. The one at CME did not.
What it means for a miner
None of this moved what a machine earns. Hashprice, a day's mining revenue per unit of hashrate, was $41.31 per petahash on 22 September, the day funding peaked, and $40.17 on 28 September, after four days of flat-to-negative funding and a futures book that was 8% smaller. The difference is under 3%, smaller than the swing in funding alone.
That is the distinction worth keeping straight, on this rally and the next one. A squeeze is leveraged positioning, and leveraged positioning reverses on its own clock — this one took four days to go from the strongest funding reading of the whole run to negative. What it left behind, the block reward paid every ten minutes and the bitcoin already sitting in a wallet, does not answer to a margin call. A miner who kept the machines running through the loud week and the quiet one after it earned within a few dollars of the same hashprice either way.
Sources and further reading
- Bitcoin long liquidations hit $280M as BTC price dips under $84K — Cointelegraph, 23 Sep 2026
- Bitcoin price nears $84K as ETF inflows reach five days — crypto.news, 24 Sep 2026
Disclosure
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.