Bitcoin: Cheap or Expensive... and What Comes Next?
How to use this chart
Four zones, one question: where in its 15-year band is BTC sitting right now? Accumulation (cheap), Trend (fair), Distribution (rich), Mean-reversion risk (stretched). The current read is at the top of the page.
Accumulation — below Q25. Multi-week stays here have always resolved one way: back above Q50.
Trend — Q25–Q75. About half of Bitcoin's life. Useful for context, less useful for timing.
Distribution — Q75–Q95. Every prior visit has eventually mean-reverted toward Q50. Sometimes after pushing higher first.
Mean-reversion risk — above Q95. About 5% of history. The model can't tell you how high a spike can go — only that it doesn't last.
Past behaviour, not prediction. A curve fit, not a causal theory. The FAQ lists what would break it.
How the chart is built
A log-log regression of BTC price against block height, since 2010. Block height removes mining-rate drift; the model converts back to calendar time for display.
The bands are empirical quantiles around that trend. Q50 is the median — half of Bitcoin's history above, half below. Q25–Q75 holds the middle 50%. Q5 and Q95 are the historic tails. Q1 and Q99 are the rare extremes the model has nearly never visited.
How the forecast band works
The dashed line and shaded cone are a conditional projection, not an extension of the trend line. The model finds every prior day BTC sat at roughly the same point in its band — the same quantile, give or take a couple of points — then reads off where those days actually went over the following weeks and months: the median path plus the spread around it (the shaded band is the 10th–90th percentile, the faint halo the 5th–95th).
Comparable days are weighted by how similar their backdrop was at the time — whether price sat above or below its 110-day average — so closer analogues count for more. It uses only outcomes that had already happened by each historical date, so there's no hindsight, and the band stops where genuinely comparable history runs thin. The read at the top of the page reports this band's median as an annualised rate (CAGR) at 90 days and one year.
It is still not a prediction. The bands describe how BTC has traded, not where it will go; both they and the cone could break down if Bitcoin’s adoption growth rate shifts, or if a single buyer becomes large enough to bend the curve.
FAQ
What is the Bitcoin power law?
A regression that fits BTC's price as a power function of block height since 2010. Long-term growth as a slow decelerating curve, not an exponential. Popularised by physicist Giovanni Santostasi.
Why fit on block height instead of calendar time?
Block production drifts with hash rate and difficulty adjustments. Fitting on block height removes that drift and tightens the bands; the model converts back to calendar time for display. It's also what separates this from the rainbow chart.
How accurate has it been historically?
Most of BTC's daily closes since 2010 have sat between Q5 and Q95 — a band ~3× wide in dollar terms, tight in log space. The forecast band above has contained realised price in —% of historical 90-day windows. Bull peaks and bear bottoms still escape it. As a long-run anchor it has held for 15 years. Stock-to-flow has not.
Can the power law predict future price?
No model can. What this one does: anchor expectations. The bands tell you how stretched price is against trend; the forward cone tells you what comparable history did from here. Useful for sizing decisions, useless for tomorrow's close. Not a trading signal.
How is this different from the rainbow chart or stock-to-flow?
The rainbow chart is one OLS line on calendar time, with subjectively-named colour bands. This is an empirical quantile regression on block height — each band is where price has actually sat. Stock-to-flow tied price to BTC's supply schedule alone; its projection diverged from realised price after 2022. The power law has not.
What is the oscillator strip beneath the chart?
The same information from another angle. Instead of plotting price next to the bands, it plots BTC's current quantile position as a single line, from 0 (Q1) to 1 (Q99). The main chart is logarithmic and slow; the oscillator amplifies short moves and makes regime crossings obvious. The strip's height at any point is BTC's quantile that day.
How long has BTC historically stayed in each band?
Typical visits are days, not weeks. Median dwell before crossing a band boundary: 5 days in Accumulation, 5 in Trend, 4 in Distribution, 5 in Mean-reversion.
Cumulative share of life: 29% Accumulation, 46% Trend, 20% Distribution, 5% Mean-reversion.
Longest streaks: 451 days in Accumulation (the 2015 bottom), 88 days in Mean-reversion (late-2013). The outliers are the cycles.
How often does the model refit?
Weekly. The historical price series, quantile fit, and forecast analogy weights all update Monday morning UTC. The calibration backtest re-runs quarterly.
What's the 110-day average referenced in the methodology?
A simple moving average of BTC's daily close over the prior 110 trading days. The forecast model uses it as a regime indicator — analogues from days where price sat above their 110d MA are weighted toward the price-above-MA case, and vice versa. A short-horizon proxy for "what kind of tape was Bitcoin in".
What to look at next
The power law tells you where BTC sits. The rest of mNAV tells you who's buying it at this price. The BTC Stacking Race tracks who's accumulating fastest across 19 listed treasury companies. BTC Stacking Curve overlays share price against BTC buys across 110+ tickers. Strategy ATM is the live model of Strategy's five ATM programs (MSTR + STRC + STRF + STRK + STRD).
For preferreds, the macro fair-value model, momentum portfolio, and the rest — open the mNAV menu at the top of this page.