This cycle's bottom
Reached 18 Nov 2026July 2026 to January 2027
ReachesCheap
Closed at $62,897
A rainbow drawn from Bitcoin's own cycles: a curve fitted to the midpoint of every bottom and top since 2010, and 9 bands cut from how far the price has strayed from it.
Each cycle gives one point, halfway between its bottom and its top on a log scale. The curve is fitted to those 5 points — the cycle still under way counts for less, since its bottom is not in yet — and it bends, because Bitcoin's growth slows decade by decade.
The width comes from the cycles too: how far each one swung around its midpoint. Those swings shrink from cycle to cycle, so the rainbow narrows going forward instead of fanning out.
The bands are cut from every daily close since 2010. The cheapest holds 6.0% of the days, the dearest 3.0%, and the 7 between them are evenly spaced. Fair price is the middle band.
It describes where the price has been against its own cycles, not where it will go. Every new cycle moves the curve.
$60,123
−52%
Counted from the day the price entered the band, today's crossing included. A change only sticks once 5 closes in a row agree: if the price goes back before then, the days away are added to the band it returns to rather than resetting its count.
Cheap · 4 days
The band the price was in before the current one, and how long it stayed. If it sits below, the price arrived rising; if above, falling. A stay only ends once 5 closes in a row agree on another band.
Slight discount · 20 days
Bitcoin's price on a logarithmic scale over nine valuation bands, from heavily undervalued to overheated, with the model's fair-price line through the middle
| Cycle | Bottom | Top | Travel |
|---|---|---|---|
| 2011 | — | +4Peak bubble | — |
| 2011 → 2013 | −3Opportunity | +4Peak bubble | 7 |
| 2015 → 2017 | −3Opportunity | +4Peak bubble | 7 |
| 2018 → 2021 | −3Opportunity | +4Peak bubble | 7 |
| 2022 → 2025 | −3Opportunity | +2Expensive | 5 |
| 2026 → 2029 | −1*Slight disc. | —pending | — |
* The cycle is still open.
Reached when a daily close goes two bands from fair price on its own day, the shortest distance any cycle has covered. The band moves with time, so the price under it holds for one day: today's while the window is open, the opening day's while it is ahead. The edges come from a fit on the 4 closed cycles only, so the open cycle's lowest close is not an input. Against the chart's bands: −0.28% (bottom), +10% (top).
July 2026 to January 2027
ReachesCheap
Closed at $62,897
June to December 2029
ReachesExpensive
Above $226,243
What buying a fixed amount regularly would be worth today.
Bitcoin measured against gold, stocks and inflation.
Sats to dollars and dollars to sats at the live price.
What the halving does to supply, and every cycle measured.
A logarithmic chart of Bitcoin's price, overlaid with colour bands that classify the market from "heavily undervalued" to "overheated" against its long-term trend. Unlike the classic static model, this one fits a power law with dampening over complete cycles and derives its nine bands from the percentiles of the price's own history.
The centre is a regression on the logarithm of days since the genesis block, passing through each cycle's midpoint — the geometric mean of its top and its bottom. Around it, nine bands are cut at percentiles of the price's historical distance to that centre. The model refits with every data update instead of staying fixed.
The “Current band” card at the top of the page shows it against the live price: from “Bargain” to “Peak bubble”, along with the share of history the price spent in that band. The band says where the price sits against its trend, not what it will do tomorrow.
The classic rainbow chart used a fixed logarithmic regression, and in 2022 the price fell out of it from below: it had to be redrawn. This model corrects that defect by measuring the deceleration instead of assuming it and refitting to the cycles — but it still describes the past: four completed cycles are only four observations, and no band guarantees the future.
Because we do not maximise it. Least squares over every day — the usual approach — maximises R² by construction: replicated on our own data it scores 96%. Our curve is anchored to the midpoint of each cycle instead. That costs a few points of R² and buys what R² cannot see: a curve that is not tilted against the cycles, the defect that left the classic rainbows projecting a top far below their own bands. R² measures average closeness to the past, not whether the cycle structure is right.
The middle of the "Fair price" band — the centre of the middle band of the nine, drawn as the dashed line. Because the bands divide the price range evenly while the price spends more of its time low in that range, it sits where 61% of history has been cheaper, not half of it — hovering any day on the chart gives that day's own percentile. It is not a target value or a prediction.
The "Model projections" card gives both: a window from the closed cycles' median timing, and the band edge the price would reach if it moves as far from fair price as the shortest distance any cycle has covered. Each carries a status, rechecked every day. The table above it shows the band every past bottom and top reached.
No. The bands describe where the price has been, not where it will go, and the curve moves with every new cycle. Use it as historical context for reading the market, never as a recommendation to buy or sell.