Bitcoin Has ‘Never’ Had a Green September After a Green August. In 2012 It Did — and the Viral Version of That Statistic Also Loses the Biggest Green August on Record

Bitcoin finishes August on Monday up 24.45% for the month, the second-largest August in its recorded history behind only 2017, and it does so with the Federal Reserve’s September meeting priced as close to a coin flip as a market can get. That combination has produced the single most-shared bitcoin statistic of the weekend: bitcoin has never had a green September following a green August — not once. The line was picked up on Saturday by The Crypto Times from a widely circulated social post, and by Sunday morning it had been reposted, screenshotted and turned into a trading thesis across the market. We pulled the daily candles and checked it. It is not true.

The counterexample is 2012. On Bitstamp BTC/USD daily data — our own direct pull, 31 daily rows for August and 30 for September, no partial months, no gaps — August 2012 returned +9.48% and September 2012 returned +19.80%. Green August, green September, and the September was the larger of the two by more than double. The figures match the independent monthly heatmap published by Newhedge to the second decimal place. The reason almost nobody has this cell in front of them is mundane and worth stating plainly: the monthly-returns table that crypto media quotes for seasonality starts in 2013. The claim is not a discovery about bitcoin. It is a description of where a spreadsheet begins.

Every green August, and what September did next

Bitstamp BTC/USD, first-of-month open to last-of-month close, UTC. Own pull, 30 August 2026, 06:20 UTC. August 2011 is excluded: Bitstamp’s series begins on 18 August 2011, making that month 14 rows rather than 31. August 2026 is measured through 30 August and is not yet final.

So the honest version of the statistic is one in five, not zero in four. Five completed green Augusts in bitcoin’s recorded history; one of them was followed by a green September. That is a 20% base rate on a sample of five, which is a very different sentence from “never, not once” — and it is a sentence no one would build a position on, which is precisely why the stronger version travels further.

The second error is bigger than the first

There is a version of this claim circulating that lists the green Augusts as “2013, 2020 and 2021”. That list is missing a year, and it is missing the most important one. August 2017 returned +65.78% — the largest August bitcoin has ever recorded, by a factor of two over the next best. Dropping it does not change the direction of the conclusion, because September 2017 fell 8.62%, but it does something worse than that: it removes the observation that most resembles the present. August 2026, at +24.45%, is the second-largest green August on record. Any conditional pattern built on green Augusts is a pattern about outliers, and the sample of outliers has exactly two members — 2017 and 2026 — one of which has not finished.

One further note on where the number comes from. The figure of +26.07% for August month-to-date has been carried in coverage since Saturday. It is not a different index; it is a stale clock. Working backwards from the 1 August open of $62,887.88, +26.07% implies a spot price of $79,282 — which Binance BTCUSDT printed at 12:00 UTC on 27 August and again at 09:00 UTC on 28 August, not on the 29th. The correct month-to-date figure at press time is +24.45% on Bitstamp and +24.32% on Binance, with Newhedge’s independent heatmap at +24.49%.

What the record actually says about September

September’s reputation is real but it is much narrower than the reputation suggests. Across the thirteen complete Septembers from 2013 to 2025 our own pull returns a mean of −3.04%, a median of −3.11%, and five positive months out of thirteen. That mean sits comfortably inside the range of published figures — CoinGlass’s since-2013 average of −3.77%, Bitcoin Suisse’s −4.16%, and Coinbase Research’s “around −3% over the last ten years” — so there is no dispute about the arithmetic. The dispute is about the shape.

The entire reputation rests on one six-year block. September fell in 2017 (−8.62%), 2018 (−5.96%), 2019 (−13.50%), 2020 (−7.55%), 2021 (−6.97%) and 2022 (−3.11%) — six consecutive negative Septembers, which is where the folklore was manufactured. It is bracketed by green on both sides. September 2015 rose 2.58% and September 2016 rose 6.57% before the streak; September 2023 rose 4.01%, September 2024 rose 7.35% and September 2025 rose 5.35% after it. Take the six-year block out and the remaining seven Septembers average +0.89%, with five of the seven positive; the block itself averages −7.62%. The “September effect” is not a property of the calendar. It is a property of 2017 to 2022.

Bitstamp BTC/USD, 2013–2025 complete calendar months, own computation. 2026 excluded throughout.

On two measures out of three, August is the worse month

This is the part of the seasonality conversation that almost never gets printed. August’s median return, −8.31%, is more than twice as negative as September’s −3.11%. August was positive in 4 of 13 years; September in 5 of 13. On median and on hit rate — two of the three summary statistics anyone actually uses — August is bitcoin’s worst month, not September. September wins the title only on the mean, and it wins there because August’s mean is dragged up by two enormous outliers in a way its median is not. Remove August 2013 (+30.92%) and August 2017 (+65.78%) and the remaining eleven Augusts average −7.61% — almost exactly the −7.62% of September’s notorious 2017–2022 block. Two months, two reputations, the same underlying number, and the difference between them is which two cells you are allowed to delete.

Which sets up the irony sitting on top of tomorrow’s close. Bitcoin has just had a +24.45% month in the month with the worst median return in its history, and the market’s response is to brace for the month with the second-worst. The same statistical apparatus that says to fear September said, four weeks ago, to fear August. It was wrong by twenty-four percentage points.

Where the ‘wrong in 2023 and 2024’ line came from

A related formulation keeps appearing: that the September trade “would have been wrong in both 2023 and 2024.” We traced it. It comes from Coinbase Institutional’s research note Monthly Outlook: Debunking the Seasonality Myth, published in September 2025 — which is to say, published during the September it could not yet count. The note was accurate on its own date. Quoted in 2026 without amendment it understates its own case, because September 2025 went on to return +5.35% and the correct count is now three consecutive years, not two. In fairness to this year’s coverage, we looked for 2026-dated pieces still using the two-year version and did not find one: AMBCrypto on 28 August, for instance, correctly lists 2023, 2024 and 2025. The stale phrasing survives in social summaries rather than in the trade press.

One editorial note on sourcing, since we are asking readers to discount a viral claim. The social account that originated the green-August formulation was publicly accused by the on-chain investigator ZachXBT in October 2024 of operating multiple accounts and deploying memecoin scams. We are not reproducing the post, and the provenance is not the reason the claim fails — it fails on the 2012 data, which would be equally decisive if a central bank had published it. But readers deciding how much weight to give a statistic should know how far it is from a primary source.

What is different this September, and it is not the calendar

The variables that will actually decide the month are all dated and none of them are seasonal. The FOMC meets on 15–16 September, with the decision at 2:00 p.m. ET on Wednesday the 16th and a Summary of Economic Projections attached. After Chair Kevin Warsh’s Jackson Hole keynote on Friday, CME FedWatch put the probability of a September hike at 55.7%, up from 35.4% on Thursday; Kalshi had it near 48%, and Benzinga carried a 59% print timestamped 12:09 p.m. ET. The spread between those three is itself the story: nobody knows. July PCE, released on 26 August, ran at 3.7% headline and 3.3% core year on year, both a tenth above consensus, and Warsh told Jackson Hole that the six-month annualised pace is 4.1%.

Against that, the flow picture for August was the strongest of the year until Friday broke it. US spot bitcoin ETFs took in roughly $3.32 billion across the month through 28 August on Farside Investors data — the best month of 2026 — before −$201.9 million on Friday ended a nine-session inflow streak worth $3.04 billion. Strategy has not bought a bitcoin in 69 days, its last purchase being 520 BTC at about $67,068 on 22 June, and has sold roughly 6,916 coins since; its next 8-K is expected on Monday. At 06:20 UTC on Sunday bitcoin traded at $78,120, up 0.71% on 24 hours, market capitalisation $1.568 trillion (own CoinGecko pull), which CoinDesk corroborated at $78,134.65 at 1:48 a.m. ET. Ether was $2,455.42, Solana $105.05 and XRP $1.39.

The honest bottom line

We are not arguing that September will be green. We are arguing that the specific statistic being used to argue it will be red is wrong on its own terms twice over — once on 2012, once on 2017 — and that even repaired it rests on five observations, one of which is unfinished. A four-out-of-five hit rate on a sample of five has a 95% confidence interval that spans almost the entire probability space. That is not a signal. It is a coin that has been flipped five times.

The useful discipline here is the one we apply to every streak on this site: print the denominator, print the window, and say what would falsify the claim. For this one, falsification is cheap and it is dated. If bitcoin closes September 2026 above its 31 August close, the pattern is 4 of 6 rather than 4 of 5, and a statistic that was being described on Saturday as an iron law will have failed twice in fourteen years. Our accompanying field guide today sets out the six checks to run on any seasonality number before it is allowed to influence a decision.

Method: monthly returns in this article are computed by Bitcoin Mastery as first-of-month open to last-of-month close, UTC, on Bitstamp BTC/USD daily candles — 4,990 rows from 1 January 2013 to 30 August 2026, plus a separate extension pull covering August 2011 to December 2013. Partial months are excluded and identified. Prices, flows, funding, open interest and on-chain figures are pulled directly at the timestamp stated. Where a third-party figure is cited we name the source and its date; where two sources disagree we print both.

Disclaimer: This article is for informational purposes only and does not constitute investment advice. Cryptocurrencies are volatile and you can lose money. Historical seasonality describes the past; it does not forecast the future. Do your own research and consult a licensed financial advisor before making investment decisions.