Rainbow

Bitcoin's price on a log scale against a logarithmic regression, with coloured bands around it. Where the price sits in the bands is the whole point of the chart.

Right now: Still cheap — 0.36 standard deviations from the fitted line.

The regression is fitted here, at build time, to the same price series the rest of the site plots — not to constants copied from another chart. The model is log10(price) = 1.4733 · ln(days) − 8.0566, least squares over 5,889 daily closes from 2010-07-17 to 2026-09-06, R² = 0.910. The bands sit at multiples of the residual standard deviation (σ = 0.462 in log10 terms), so their width is a property of how far the price has actually strayed from the fit rather than a number someone picked.

What it is not: a forecast. A log regression through a rising series will always slope up, and the band you are in says where the price is relative to its own past, not where it is going. Refreshing the dataset refits the line and moves every band.