The hard questions about this chart. Answered with our own numbers.
Counted from our own data
648 days
closed outside the bands entirely
591 days below the bottom band. 57 days above the top.
Every daily close on record: 5,921 days, Aug 18, 2010 to Nov 2, 2026. And the line was drawn with hindsight.
The line is a curve fit to the past. It is not a law. No force makes the price follow it.
And it was fit knowing the ending. Fit the same line using only prices through 2013, then judge the years after. The price spent 3,249 of the next 4,689 days below the whole ladder. That is 69% of the time. Drawn in 2013, the chart failed. Drawn today, it hugs. The difference is hindsight.
The 0.96 fit score flatters it, too. On log scales, any price that rose a lot scores high. Judge by distance instead: half of all days closed at least 1.65 times away from the line, over or under. Only 530 of 5,921 days closed within 10% of it.
The chart is also silent where people ask most. It says nothing about why the price moved. It says nothing about next month.
One rule sits under the chart. The line rises with age to the power 5.6. Each time bitcoin’s age doubles, the line climbs about 48 times.
Early on, age doubled in months, so the line rockets. Now an age doubling takes about 18 years, so the same rule crawls. Same rule, slower clock. That is the bend.
The math way, the curve is not a curve. Plot price in logs against time in logs, and it runs straight. Spread the years out evenly, as charts do, and the straight line bends.
The bands are the line times fixed steps. Each band is 1.35 times the band below it. So the bands bend with the line, and hold the same width in percent.
Yes: 648 of 5,921 days, the count in the card above.
Above the top: 57 days, all early. In June and July 2011 the price lived over the top for 27 days. On June 11, 2011 it closed at $33.80, with the top of the ladder at $14: 2.3 times the top. In late 2013 it escaped again, 30 days, peaking at $1,134 against a $729 top on Nov 30, 2013. No day has closed above the top band since Jan 13, 2014.
Below the bottom: 591 days. The price lived under the ladder for 309 days across 2010 to 2013, and 240 more across 2015 and 2016. The longest stretch out ran 89 days, Oct 25, 2012 to Jan 21, 2013. The deepest day: Oct 7, 2010, a 6-cent close at 38% of the bottom edge.
Since then, only dips. One day in the March 2020 crash: $4,830 against a $4,944 bottom edge. Ten days at the November 2022 bottom, and 26 more that December and January, never 5% under the edge. Five days in June and July 2026, a hair under.
Counted on this site’s fit. Other hosts draw different bands, and their charts break in different places. That story comes next.
The famous version lives at Blockchain Center. Its keeper, Rohmeo, says it plainly: the chart was “NEVER a serious attempt to model or predict the price of Bitcoin.” His words: a meme, a joke, with “no scientific basis.”
It started as two ideas that met. In 2014, a Reddit user, azop, drew rainbow bands over the price to cheer people up in a crash. That October, a forum user, trolololo, fit the growth line under it. Rohmeo merged the two in 2019. He says he set the band widths and labels by hand.
Yes, and its keeper moved it. All of this is the Blockchain Center chart. Its bands are not this site’s bands.
The price closed over its top in November and December 2013. It dipped under its bottom in the March 2020 crash. In November 2022 it broke under for real, bottoming near $15,500 to $15,800 as FTX fell.
Rohmeo first added a new band under the price. On Nov 21, 2022, he published a full refit, V2: the 2014 formula had grown too rosy. Today’s version, called Dynamic, refits itself as new prices land.
A critic at LUV names the worry: the fit “is refitted every time new data arrives, so the curve quietly moves to keep explaining whatever just happened.”
There is no one rainbow, either. Blockchain Center draws 10 bands. CoinGlass draws 9. CoinGecko, in 2026: “There is no industry wide accepted formula.”
Ours differs in one way that matters. This site prints its exact formula, its version, its data window, and its data seam on the method page. Anyone can redo the math. The line is refit over the whole record as new days land. The formula stays the same. It is not swapped for a new one when the old one breaks.
The math has a name. In January 2024, Giovanni Santostasi showed the rainbow’s regression is a power law in disguise: price growing as a power of age. The same family as the line on this site.
The sharpest attack came earlier. Marcel Burger, in 2019, on the power-law corridor: the famous fit score “is not a very meaningful metric” when both series trend. The match can be an accident of two lines rising together. The fit’s leftovers move in long waves, not random scatter. And day zero, the genesis block, is a cherry-picked start. His verdict: the model “is falsified.”
Defenders answer back. They point to a mechanism: adoption grows, and value follows the network, Metcalfe-style. And they say the fit’s settings have barely moved since about 2017. The honest state of it: unsettled. Serious people hold both sides.
No one has scored the rainbow the fair way: as a buy-and-sell signal, on years the bands were not drawn through, against a plain yardstick like the 200-week average. That test does not exist.
One backtest does. A “rainbow-weighted” plan bought more in low bands and less in high ones, and beat plain steady buying on paper. But it was scored on the same history the bands were fitted to. Its replicator could not find the original band settings, and invented his own. And the winning plan put in more cash than the plan it beat.
Even the friendly guides land in the same place. Binance Academy and TokenTax both frame the chart as a context lens, one input among several. Never a signal on its own.
No. It is a ruler laid over history. It says where the price sits inside its own past. Nothing more.
Being low in the band ladder has not reliably predicted what came next. A low band is a fact about today, read against the past. It is not a fact about next month.
Our line is refit over every close on record each time the data rebuilds. With 5,921 days behind it, one new day barely moves it. So a break cannot hide: the line will not quietly chase the price.
A break looks like this. The price lives outside the ladder for months, not days. Or years pass with growth nowhere near the fitted slope.
If that comes, the honest move is to say so. The fit score on the method page would fall. The bands would no longer hold the past they were drawn from.
Every number on this page is counted from the same record as the method page: daily closes, Aug 18, 2010 to Nov 2, 2026.