Why So Many Bitcoin Indicators Have Stopped Working

Work through this site's charts in order and the same thing keeps happening. An indicator that called the 2013 and 2017 cycle tops precisely does nothing at the 2021 and 2025 ones. A threshold that marked every previous peak is never reached again. A fitted band that price used to exceed is now a long way overhead.

This is not a coincidence across unrelated indicators. It is one effect showing up in several places, and it has a straightforward cause.

The evidence

Pi Cycle Top has fired four times in the entire record: April 2013, one day after the December 2013 peak, on the exact day of the December 2017 peak, and in April 2021 before that year's spring high. The November 2021 cycle top produced no crossover — the fast average sat at $49,510 against $87,260 for the slow one, nowhere near. The October 2025 top was further away still: $113,700 against $199,800. Two exact calls, then two cycle peaks missed entirely.

The power-law rainbow has had price close above its top band on 151 of 4,345 days, and the last of those was 2021-05-11. At the 2017 peak price stood 2.3× above that band. At the 2021 peak it was just under. At the 2025 peak it reached 38% of it.

MVRV peaked at 4.72, 4.43, 2.86 and 2.29 across the four cycles. The conventional 3.7 line has not been touched at either of the last two tops. NUPL, being the same arithmetic, shows the same decay: 0.79, 0.77, 0.65, 0.56, with its 0.75 euphoria threshold last reached in 2013.

Long-Term SOPR read 16.1 and 20.6 at the first two cycle tops and 2.9 and 2.4 at the last two.

The Puell Multiple read 7.25 and 6.62, then 1.64 and 1.16.

One cause, several symptoms

Every one of those indicators is, directly or indirectly, a measure of how steeply price has risen. Pi Cycle needs a fast average to outrun a slow one by a factor of two. The rainbow needs price to exceed a trend fitted through Bitcoin's early years. MVRV needs price to run far ahead of what holders paid. SOPR needs patient holders to be sitting on large multiples.

Bitcoin no longer rises that steeply. Peak-to-peak gains have gone 17.3×, then 3.4×, then 1.9× — each cycle roughly a fifth of the last. Realized volatility has fallen from a mean of 110% in 2013 to 41% in 2025, and the October 2025 peak arrived with a 30-day reading of 27.5%. An indicator calibrated on an asset that routinely tripled in three months does not trigger on an asset that gains 90% in eighteen.

There is a second mechanism behind the cost-basis metrics specifically. Each cycle draws in a larger base of holders at higher prices, so the aggregate cost basis climbs faster and the ratio has less room to stretch. MVRV's ceiling falls for the same reason its denominator grows.

What has survived

The decay is almost entirely on the upside. Every one of these metrics has kept its bottom-side signal:

MVRV below 1 at every major low — 0.56, 0.69, 0.75
NUPL negative at all three, meaning the market as a whole held a loss
Long-Term SOPR at 0.50, 0.42, 0.44 — barely moved across twelve years
Short-term holders' realized price falling below long-term holders', four times, each inside a bottom

The asymmetry is not an accident. Upside readings depend on how large the holder base's gains are, which shrinks as an asset matures. Downside readings depend on what people do under loss, and a year-long holder selling at a loss is the same act of capitulation in 2015, 2018 and 2022. Human behaviour under pressure has a floor that arithmetic does not.

What to do about it

Distrust absolute thresholds. Any number that was set by looking at 2013 and 2017 has already failed twice. Measures expressed in relative terms — standard deviations from a mean, or distance from a trailing average that adapts — degrade more slowly, though the z-score has its own version of the problem, since every new day of high valuations widens the distribution it grades against.

Prefer trailing measures to extrapolated ones. The 200-week moving average has tracked price through the last two cycles because it follows the market. The power law and the rainbow have drifted away from it because they extrapolate a slope from the steep early years.

Weight the bottom signals. They have held. The top signals have not.

Assume this page ages too. If the mechanism above is right — that the decay comes from maturation — then the metrics that still work are the ones that have not yet been asked to. It would be perfectly consistent for the bottom-side signals to lose their edge as the holder base changes further, and nothing here would have predicted it.

Why we publish the failures

An indicator that worked twice and then stopped is more informative than one presented as though it never stopped. Every metric page on this site now states what its own record actually is, including where it contradicts the indicator's reputation, because a chart tuned until it agrees with the past tells you about the tuning rather than about the market.

Last updated 2026-08-10.