Does the full moon move Bitcoin?

The short answer

No. The interesting part is not the answer but what it takes to get it right.

What we measured

61 full moons over five years, with timings computed astronomically rather than from the average lunar month. That approximation drifts by up to fourteen hours, which is enough to put an event on the wrong day.

Bitcoin, Ethereum, Solana and XRP were weighted into one observation per full moon. Four correlated coins are not four independent measurements; treating them as such would have quadrupled the apparent sample without adding a single fact.

The control decides it

Bitcoin fell into 24 of the full moons in the window, and 12 of those rose over the following week. Fifty per cent — a coin flip.

But the number that settles it is the control: the same decline on days that were not full moons did better. The moon condition performed worse than the absence of the moon. Whatever mild mean-reversion follows a fall, it happens on ordinary days too.

Distance and duration are one question, not two

A natural follow-up is whether a supermoon differs, or whether it matters how long the moon appears full. These turn out to be the same variable: the moon travels faster near perigee, so a close full moon is also a short one. The two measures correlate at 0,94.

Neither relates to what happens afterwards. Testing both as if they were independent would have been counting one test twice.

What is still open

One narrower version has not been dismissed. Conditioned on a small decline into the full moon rather than a large one, the sample turns positive and holds out of sample. But it rests on thirteen observations, the boundary was chosen after seeing the data, and thirteen is below the floor we require before calling anything a finding.

It is registered, pre-specified, and being measured forward. If it reaches a real sample and still holds, we will say so here. If it does not, this page will say that instead.