Pi Cycle Clears Its Own Line by Less Every Cycle

The Pi Cycle Top indicator fires when Bitcoin's 111-day moving average crosses above twice its 350-day moving average. It has a reputation for landing close to cycle peaks. We recomputed it from our own price history, and the interesting part is not that it worked — it is how much less decisively it has cleared its own threshold each time.

Signals found: 3 · 2013 margin: 21.6% clear · 2017 margin: 5.1% clear · April 2021 (weekly): 0.99998 of the line · Now: 56% below the line

The construction

Two moving averages, one doubled. The 111-day average tracks the recent trend; the 350-day average tracks the cycle. Doubling the slow one creates a ceiling that the fast one only reaches during a violent, sustained run — which is what a blow-off top looks like from the inside.

The numbers 111 and 350 are chosen partly because 350 divided by 111 is close to pi, which is where the name comes from. That is a coincidence dressed as a rationale, and it is fair to say so: nothing about circles is involved. What the pair does have is a wide separation between a fast and a slow window, which is what makes the crossing rare.

What our own series shows

Reconstructed at weekly resolution across 845 sampled weeks from July 2010, the fast line crossed above the doubled slow line three times: the week of 8 April 2013 with Bitcoin at $162, the week of 28 November 2013 at $1,080, and the week of 14 December 2017 at $16,224. Each landed close to a cycle peak.

The revealing figure is not the date but the margin — how far past its own threshold the indicator went. In April 2013 the fast line peaked at 1.216 times the doubled slow line, clearing the threshold by 21.6%. In November 2013 it reached 1.168. In December 2017, 1.051. The margin has fallen every cycle: 21.6%, then 16.8%, then 5.1%.

In April 2021 our reconstruction shows it reaching the line and not passing it. At the closest point, the week of 8 April, the fast line stood at $46,789 against a threshold of $46,790 — a ratio of 0.99998. That is a graze at weekly resolution, and the next section is about why that is a weaker statement than it looks.

The margin has fallen every cycle: 21.6%, then 16.8%, then 5.1%.

What that one dollar does and does not prove

It would be easy to write that the 2021 signal never happened. That would be an overstatement, and the reason is sampling. Our price source publishes roughly one close per week before 2022, so this reconstruction is weekly. The indicator's publishers report a daily crossing in April 2021, and a crossing that opens and closes inside one week is invisible at weekly resolution.

So the honest statement is narrower and, we think, more useful: at weekly resolution the 2021 signal is a graze rather than a crossing, and whether it fired at all depends on a sampling choice most readers never see. A signal that survives only at one resolution is a fragile signal, whatever it did afterwards.

The trend across the three confirmed crossings points the same way without needing that argument. A threshold cleared by 21.6%, then 16.8%, then 5.1%, is a threshold being approached ever more narrowly.

Why the margin is shrinking

There is a mechanical explanation that does not require the indicator to be broken. The crossing needs the 111-day average to double the 350-day average, which requires the price to roughly double over a few months and hold there. As Bitcoin's market capitalisation has grown, moves of that size have become rarer, so the fast line has less room to run away from the slow one.

If that is what is happening, the indicator has not stopped working so much as outlived its calibration — the same thing the Mayer Multiple's 2.4 threshold appears to have done. We cannot establish that from three observations, and three observations is all any cycle-based Bitcoin indicator has. That is worth sitting with before treating any of them as a rule.

Move the windows slightly and the record changes

If the pi story were doing real work, the result should not depend much on the exact windows. We tested that by running the same rule across nine nearby fast/slow pairs on the same weekly-sampled series.

The signal count ranges from one to five. A 20-week/50-week pair fires once, in 2013. The published 16/50 pair — the weekly stand-in for 111/350 — fires three times. A 13-week/50-week pair fires five times. Moving the fast window by three weeks in either direction changes the answer.

One pair is worth singling out. At 15 weeks and 48 weeks — roughly 105 and 336 days, a nudge from the published numbers — the rule does produce a signal in the week of 8 April 2021, exactly where the published pair falls a fraction short. The 2021 question turns on window lengths chosen for their resemblance to pi.

Three observations is not a track record

Every cycle-based Bitcoin indicator shares one problem, and it is not a subtle one: Bitcoin has had three completed cycles. Any rule tuned to fire near a top has at most three chances to be judged, and a rule with three successes and no failures is indistinguishable, statistically, from a rule that got lucky.

It is worse than that for a threshold rule, because the threshold itself was chosen after seeing the data. Picking 111 and 350 because those windows separate neatly on the history you already have is curve-fitting, however reasonable the pi story sounds. The test that would mean something is out-of-sample performance, and there have been three opportunities in sixteen years.

None of this makes the indicator useless. It makes it a description of what happened three times, which is a different object from a forecast, and the shrinking margin is a reason to hold even the description loosely.

Where it stands today

The 111-day-equivalent average is around $68,550. Twice the 350-day-equivalent average is around $157,342. The ratio is 0.44, so the fast line would need to rise about 130% relative to the slow one before the indicator had anything to say.

In other words it is nowhere near a signal, and it would take a violent, sustained run to get there. That is the normal state of this indicator: it is quiet almost all of the time, which is both its appeal and the reason its track record is three data points long.

Frequently asked questions

Did Pi Cycle call the 2021 top? Its publishers say yes, on daily data, in mid-April 2021. Our weekly-sampled reconstruction puts that week at 0.99998 of the threshold — just under. Both statements can be true; they are measuring at different resolutions.

Why 111 and 350? Because 350 divided by 111 approximates pi. That is the whole justification, and it does not survive much poking: nine nearby window pairs on our series produce between one and five signals, so the specific numbers are doing more work than the story admits.

Can it call bottoms? No. The construction only produces a signal when a fast average exceeds a doubled slow one, which cannot happen in a downtrend. There are separate indicators aimed at bottoms, and they have the same small-sample problem.

Pi Cycle is a good example of an indicator whose reputation has outrun its evidence. Three crossings in sixteen years on our reconstruction, each clearing its own threshold by less than the last, a fourth that grazes the line at weekly resolution, and a family of near-identical rules that disagree about how many signals there ever were. That is a record worth knowing about and far too thin to trade on, and the honest way to hold it is as a description of three moments rather than a rule about the next one.

Related reading: The Cycle That May Not Be a Cycle on how little history these rules rest on, and Why No Two Rainbow Charts Agree on fitted models that drift.

This essay is part of BTCDash Research. Figures were computed on 2026-09-22 from BTCDash’s own data. Nothing here is financial advice — it is analysis and background for your own research. Bitcoin is volatile; do your own diligence.