Is there a best month to buy bitcoin?
Across 188 closed months, no calendar month stands out from chance.
Every month since 2011
Colour is the size of the move; the figure is the real return
| Jan | Feb | Mar | Apr | May | Jun | Jul | Aug | Sept | Oct | Nov | Dec | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 2011 | +75% | +64% | −8.79% | +346% | +149% | +86% | −17% | −39% | −37% | −37% | −9.16% | +59% |
| 2012 | +17% | −12% | +0.35% | +1.07% | +4.85% | +29% | +40% | +8.38% | +23% | −10% | +13% | +7.62% |
| 2013 | +51% | +63% | +181% | +48% | −8.13% | −30% | +9.55% | +31% | −1.78% | +61% | +451% | −35% |
| 2014 | +10% | −31% | −17% | −1.42% | +40% | +1.76% | −8.92% | −18% | −19% | −13% | +12% | −15% |
| 2015 | −32% | +17% | −4.01% | −3.43% | −2.91% | +15% | +7.86% | −19% | +2.62% | +33% | +20% | +14% |
| 2016 | −15% | +19% | −4.98% | +8.16% | +18% | +27% | −7.34% | −8.41% | +6.39% | +15% | +6.20% | +31% |
| 2017 | −0.05% | +23% | −8.98% | +28% | +66% | +6.67% | +17% | +66% | −8.56% | +48% | +56% | +39% |
| 2018 | −28% | +2.56% | −33% | +33% | −19% | −15% | +21% | −9.06% | −5.99% | −4.51% | −37% | −7.20% |
| 2019 | −7.51% | +11% | +7.95% | +29% | +62% | +27% | −7.23% | −4.51% | −14% | +11% | −17% | −5.15% |
| 2020 | +31% | −8.28% | −25% | +35% | +9.03% | −3.07% | +24% | +3.00% | −7.76% | +28% | +42% | +48% |
| 2021 | +14% | +37% | +30% | −1.79% | −35% | −6.02% | +19% | +13% | −7.28% | +40% | −7.06% | −19% |
| 2022 | −17% | +12% | +5.52% | −17% | −16% | −39% | +21% | −14% | −2.94% | +5.43% | −16% | −3.80% |
| 2023 | +40% | +0.07% | +23% | +2.97% | −7.30% | +12% | −4.15% | −11% | +3.94% | +28% | +8.88% | +12% |
| 2024 | +0.88% | +44% | +16% | −15% | +11% | −6.85% | +3.07% | −8.86% | +7.29% | +11% | +37% | −3.19% |
| 2025 | +9.50% | −18% | −2.12% | +14% | +11% | +2.33% | +8.11% | −6.50% | +5.22% | −3.88% | −17% | −3.41% |
| 2026 | −10% | −15% | +1.87% | +12% | −3.54% | −20% | +7.43% | +25% | — | — | — | — |
Month by month
Each calendar month, averaged across every year
| Month | Years | Mean | Median | In green | Chance of luck |
|---|---|---|---|---|---|
| Jan | 16 | +8.74% | +5.19% | 56% | 1.00 |
| Feb | 16 | +13% | +12% | 69% | 1.00 |
| Mar | 16 | +10% | −0.88% | 50% | 1.00 |
| Apr | 16 | +32% | +10% | 69% | 1.00 |
| May | 16 | +18% | +6.94% | 56% | 1.00 |
| Jun | 16 | +5.35% | +2.05% | 56% | 1.00 |
| Jul | 16 | +8.28% | +7.99% | 69% | 1.00 |
| Aug | 16 | +0.48% | −7.45% | 38% | 1.00 |
| Sept | 15 | −3.72% | −2.94% | 40% | 1.00 |
| Oct | 15 | +14% | +11% | 67% | 1.00 |
| Nov | 15 | +36% | +8.88% | 60% | 0.80 |
| Dec | 15 | +7.87% | −3.19% | 47% | 1.00 |
"Chance of luck" is the probability of a gap this large between this month and the rest if the calendar did not matter, corrected for testing all twelve. Below 0.05 would count as a signal; none gets there.
Methodology and limits: how this test was run
How it is tested
If the calendar month did not matter, its labels would be interchangeable. So they are shuffled 20,000 times, counting how many of those random grids separate the best from the worst month as far as the real one does. If that happens often, what you are looking at is luck.
What this does NOT prove
Finding nothing is not the same as there being nothing. The best month would have to beat the worst by more than 59 percentage points of average monthly return before this sample could call it real; the real gap is 40. What can be claimed is that there is no effect of that size, not that there is none at all.
Twelve months are twelve chances
Testing twelve hypotheses at once all but guarantees one will look extreme by luck. That is why every probability in the table is multiplied by twelve (Bonferroni correction). Without that adjustment, the most extreme month would sit just short of the threshold instead of far from it.
Closed months only
A month counts once the series reaches its last calendar day. The month in progress does not: half a month presented as whole moves the figure and contaminates the test.
The colour is clipped, the figure is not
Intensity saturates at ±50% so that November 2013 (+451%) does not leave the rest of the grid in the same pale tint. Every cell still prints its real return.
Not the calendar; the cycle, a little
Grouping the same months by their position in the halving cycle rather than by the calendar, a shape does appear: up for the first eighteen months, down between 18 and 30, then recovering. It does not reach significance either with four cycles, but it is where to look.
More tools on the same data
Frequently asked questions
Seasonality, "Uptober", and what sixteen years of months can support
Is Uptober real?
Not in this data. October averages +14% since 2011, above the overall mean, but the probability of seeing that gap by luck alone is 100% once corrected for having tested all twelve months. Put another way: with twelve months, one standing out is what you expect, and October does not stand out more than chance would make it.
Is there a best month to buy bitcoin?
None this data can identify. Across 188 closed months since 2011, the gap between the best and worst calendar month is one chance produces half the time. That does not prove no effect exists: it proves that if one does, it is smaller than the available history can detect.
Why does September have a bad reputation?
Because it averages −3.72% and only 40% of its years close green — the only month with a negative mean. But that sits inside what chance produces, and its corrected probability is 100%. The reputation is real; the measurable effect is not.
So Bitcoin seasonality does not exist?
That is not what the data says, and the difference matters. What it says is that no effect large enough to be detected with this sample is present. A small, persistent bias would be invisible here. The honest move is to bound the size, not to deny the existence.
Where is there structure, then?
In the halving cycle, not the calendar. Grouping the same returns by months elapsed since the halving produces a coherent shape — climb, top, fall, recovery — which the cycle page documents in detail. It is still a pattern over four cycles, with all the uncertainty that carries.