Home / Blog / September 2026

Data insight · Sep 23, 2026 · 7 min read

Bitcoin just had the best September on record

September is supposed to be bitcoin’s worst month, and for years it was. This year it is up 10% with a week to go, the best September in the thirteen years we have prices for. Here is what the record shows, and what it meant for anyone mining through it.

Fallen red and orange maple leaves along a granite curb in low autumn sun

Where the curse comes from

Measured from the last close of August to the last close of September, bitcoin fell in seven of the thirteen Septembers from 2013 to 2025. The average September lost about 4% and the median lost 3.4%. The worst, in 2014, took a quarter off the price.

What gave the month its name was the run. From 2017 to 2022 bitcoin closed September lower six years in a row: −10.3%, −5.5%, −13.8%, −7.6%, −9.2% and −3.4%. No other month on the calendar has lost six years running; the next longest streak is five. By the end of it “September is bad for bitcoin” had become something traders said the way farmers talk about frost dates.

The streak made the reputation more than the averages did. By simple count, August has been the weaker month: it has closed lower in nine of the fourteen years since 2013.

Six losing Septembers in a row, from 2017 to 2022. That is where the reputation was earned.

Chart: Bitcoin’s September return every year since 2013, from the last close of August to the last close of September. This year’s bar runs to the latest close.

The curse had already lifted

The streak broke in 2023, and quietly: September 2023 finished up 0.4%. September 2024 gained 8.4%, then the best September on record. September 2025 added 4.5%.

This year is the fourth rising September in a row. Bitcoin closed August at $78,750 and closed at $86,683 on 22 September, up 10.1%, ahead of 2024 with a week of the month still to run. It traded above $85,000 on 21 September for the first time since January, as The Block reported.

A calendar effect that everyone knows about tends not to last. If enough traders expect September to be weak, they sell in August, and the weakness arrives early or not at all. The bigger change is who is buying. Since January 2024 a large share of new demand has come through the US spot ETFs, whose buyers add on a schedule of their own rather than on a trading desk’s seasonal chart.

This September began in August

This month’s gain cannot be separated from the one before it. Bitcoin spent late July and the first half of August between $63,000 and $65,000, with the Fear & Greed Index at 25 on the last day of July. Between 18 and 21 August it went from $64,000 to nearly $77,000, including a single day up almost 10%, and August closed up 24.3%.

September then did something Septembers rarely do: it held a big August gain. The price spent the first half of the month between $76,000 and $80,000, touched its low at $75,839 on 16 September, the day the Federal Reserve raised rates, and then climbed 14% in six days to $86,683 on 22 September. By then the Fear & Greed Index read 78, extreme greed.

Chart: Bitcoin since June, with the Fear & Greed Index overlaid. The August break-out and the September run both came with sentiment swinging from fear to greed. Open the chart

Who did the buying

Two forces carried the move, and both are on the Data section. Money went back into the US spot bitcoin ETFs: after net outflows of $4.7 billion from January to mid-August, the funds took in $5.1 billion between 17 August and 22 September, including $999 million on 21 September alone, the biggest day of the year. And traders who had bet on further falls were forced out: futures exchanges force-closed $1.36 billion of short positions on 19 August, the largest day in the six months of liquidation data we hold.

Those two stories, the flows and the squeeze, are told in full in our companion piece on the short squeezes and the ETF inflows. The backdrop they happened against, with ten-year Treasury yields at 5% and the Fed raising rates, is in the macro paradox.

What has followed a good September

In the six years September finished higher, the fourth quarter was higher five times: 2013, 2015, 2016, 2023 and 2024. The exception was last year, when a 4.5% September was followed by a 22.5% fall to the end of December. After the seven losing Septembers, the fourth quarter rose three times and fell four.

Thirteen Septembers is a small sample, and the one year that broke the pattern is the most recent. The record says a good September has usually led somewhere, and it also says the price is still about a third below the $126,198 peak of 6 October 2025.

What it meant for a miner

For anyone mining, the price is only half the story. The other half is how many machines are competing for the same blocks. Network hashrate rose only about 4% between mid-August and 22 September while the price rose 37%, so most of the gain went straight into what a unit of hashrate earns. Hashprice went from $30.50 per petahash per day on 14 August to $40.31 on 22 September, up 32%.

On our cost-to-mine chart, a current 13.5 J/TH machine on our 3.9¢ Nigerian power spent about $26,500 of electricity per bitcoin mined in late September, against a market price above $86,000. Every machine on the network earned 32% more per terahash. The difference cheap power makes is that the machine was already making money at $63,000.

The coins are the same coins in any month. For a miner who keeps what the machines produce, a strong September does not change the business. It changes what the stack is worth this morning.

A September like this one pays every machine on the network. On cheap power, the machine was paying before it started.

Sources and further reading

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.