Is Uptober Real? Every Bitcoin October Since 2011, Measured

Every year on 1 October, crypto social media fills up with one word: Uptober. The idea is simple. Bitcoin goes up in October.

You'll usually see it backed by a table of past Octobers, mostly green. So we built that table ourselves, for every month, not just October. We used every daily close of bitcoin on Bitstamp and measured each whole month from September 2011 to August 2026. That's 180 months, and each calendar month gets exactly fifteen years.

Finding 1: October's record is good

Let's give Uptober its due. Across all 180 months, bitcoin closed higher 100 times, about 56% of the time. October beat that. It was up 10 times in 15 years, about 67%.

Its median change was +11.0%, against +2.9% for a typical month. The best October was 2013, at +61%. The worst was 2011, at −32%.

Finding 2: it isn't the best month

Look along the chart and October isn't alone. July was up 11 times, one more than October. February and April were up 10 times each, the same as October. February's median, +11.1%, is a touch higher than October's.

But October is the month with the nickname, so it's the month people check.

We use the median and the count, not the average, because one wild month can wreck an average. November 2013 rose 450%. That single month lifts November's average to +37%, even though its median is +8.8%.

Finding 3: luck explains it

Here's the test the Uptober tables skip. With 12 months to choose from, one of them will look good even if the months are all the same. So how good would the luckiest month look by chance?

To find out, we took all 180 monthly changes and dealt them at random into 12 fake months of 15. Then we did it again, 20,000 times. Each time, we checked how many up years the best fake month had.

The best fake month was usually up 11 times out of 15, one more than October. It did at least as well as October in 99.8% of the shuffles. In other words, a record like October's is almost guaranteed to show up somewhere in the calendar, even when there's no pattern at all.

The same goes for the other end. September is bitcoin's weakest month, up only 6 times, with a median of −3.2%. But the worst fake month did that badly or worse in 94.8% of the shuffles.

Finding 4: it's really two records

Uptober's record isn't spread evenly. From 2011 to 2014, October was up only once in four. From 2015 to 2024 it was up nine times in ten. The only miss was 2018, at −4.5%.

Then October 2025 fell 4.0%. That was the month of the 10 October crash, when a wave of forced selling hit crypto futures.

"Red September" did the same thing in reverse. It fell six years in a row, 2017 to 2022. Then it rose in three straight years, 2023 to 2025. This September was up 6.4% through 29 September.

Streaks like these are how a month gets a reputation. They're also what luck looks like in a short record. Flip a coin ten times and there's nearly an even chance of four in a row landing the same way.

Finding 5: you can't trade it

Say you took Uptober seriously and held bitcoin only in October, every year from 2011 to 2025. $100 would have become about $525.

Holding in the other eleven months instead would have turned $100 into about $187,060. Holding the whole time turned it into about $982,250. October helped. But almost all of bitcoin's gains came in the months Uptober tells you to ignore.

How to read it

• Ask whether a random month could have done as well. With twelve to pick from, the best one always looks special.
• Ask what would cause it. Grain has a harvest and gas has winter. Bitcoin's October has no cause like that.
• Count the years, not just the green ones. Fifteen Octobers is a small sample, and the early years went the other way.
• Watch out for streaks. Six red Septembers in a row made a reputation, and then the next three were green.
• Don't time the whole year around one month. Missing the other eleven costs far more than October ever paid.

Method, and what this cannot tell you

We took every daily close of bitcoin on Bitstamp, in US dollars, from August 2011 to 29 September 2026. For each whole month from September 2011 to August 2026 we measured the change from the last close of the month before to the last close of the month. That's 180 months, fifteen of each calendar month. Then we ran a control: we dealt all 180 monthly changes at random into 12 fake months of 15, 20,000 times, and counted how often the best fake month looked as good as October.

• Fifteen Octobers is a small sample. One or two different years would change every number here.
• Closes only, in UTC. A month measured on another exchange or in another time zone starts and ends a few hours differently. Binance's candles agree with these to within 2 points on every September and October since 2017.
• Bitstamp was a small exchange in 2011-2013, and prices from those years differ from one exchange to the next.
• September 2026 wasn't finished on the study date. It's reported as "so far" and left out of every count.
• The shuffle test asks one question: could luck produce a month this good? It can't prove October has no real effect. It shows the record doesn't need one.
• Month-of-year patterns are famous, so traders act on them. That can make a pattern show up for a while, and it can make it stop.