BITCOINREVERSAL RESEARCH / Cycle indicator
Bitcoin Pi Cycle Top
Track Bitcoin’s two cycle averages and their distance from the Pi Cycle crossover.
Bitcoin Pi Cycle Top live chart
Daily closes · Drag to pan · Pinch or scroll to zoom
Original indicator: Philip Swift (2019) · Recreated by Simon · BitcoinReversal
How to interpret the Pi Cycle Top Indicator
On Sep 19, 2026, the Pi ratio is 0.427. The 111-day average ($67,666) is below the doubled 350-day threshold ($158,420). There is no new Pi top signal for this observation. The latest crossing in this series was Apr 11, 2021.
A low Pi ratio is not a buy signal. It does not mean Bitcoin is cheap or protected against a fall.
What does the Pi Cycle formula measure?
The Pi Cycle Top Indicator compares Bitcoin’s 111-day simple moving average with twice its 350-day simple moving average. The ratio is 111DMA ÷ (2 × 350DMA). Philip Swift introduced the model in April 2019. The 350 ÷ 111 relationship is approximately 3.153, close to π; it does not establish a causal law of markets.
Only the upward crossing of 1.00 defines a new signal. Remaining above 1.00 does not produce another signal every day. Weekly and monthly views sample completed daily calculations.
One signal. A broader decision framework.
Explore how BitcoinReversal brings defined market and cycle conditions into a separate, rules-based framework.
Historical testing of the Pi Cycle Top Indicator
Crossings and missed warnings belong in the same record. These are reconstructed observations from this site’s daily price series, not a historical live trading record.
3Crossings before April 2019
Already observable when the model was introduced
1Crossings after April 2019
A very small later sample
3 days before nearby high
-55.4%
Lowest close vs. signal in the next 90 days
1 day after nearby high
-50.0%
Lowest close vs. signal in the next 90 days
Same day as nearby high
-64.3%
Lowest close vs. signal in the next 90 days
2 days before nearby high
-47.4%
Lowest close vs. signal in the next 90 days
Compare exact prices and the ±30 / 90-day measurements
| Signal date | BTC close | Nearby ±30-day high | Timing vs. high | Lowest close vs. signal, next 90d |
|---|---|---|---|---|
| Before publication | $153 | $228 Apr 10, 2013 | 3 days before high | -55.4% |
| Before publication | $1,045 | $1,151 Dec 4, 2013 | 1 day after high | -50.0% |
| Before publication | $19,497 | $19,497 Dec 16, 2017 | Same day | -64.3% |
| After publication | $60,205 | $63,503 Apr 13, 2021 | 2 days before high | -47.4% |
“Nearby high” is the highest close from 30 days before to 30 days after a crossing. The 90-day statistic is the lowest close in the next 90 calendar days relative to the signal close. Both are hindsight descriptions; neither models executable returns.
Price can peak while the ratio stays low
November 2021 reached a higher close after the April signal, without a second crossing. A later example from 2025 makes the same limitation tangible.
November 2021 · no new crossing
- Selected close
- Nov 8, 2021
- BTC / USD
- $67,567
- Pi ratio
- 0.568
- Lowest close, next 90d
- -48.2%
A higher nominal close than the April 2021 high; no second upward crossing.
October 2025 · no new crossing
- Selected close
- Oct 6, 2025
- BTC / USD
- $124,753
- Pi ratio
- 0.569
- Lowest close, next 90d
- -32.1%
The highest 2025 daily close in this stored series; no upward crossings during that year.
Examples selected in hindsight from this site’s stored price series. Each card compares a selected high with the lowest close in the following 90 days. These are not investable backtests or proof of a final cycle top. Imported historical vendor provenance is unverified.
What Pi Cycle cannot establish
Future timing
A few historical crossings do not establish a stable forecasting advantage. Parameters selected from earlier cycles may lose relevance.
Fair value or a floor
The ratio measures one trend relationship. Low readings can coexist with large subsequent losses.
Executable returns
This chart includes no strategy rules, fees, slippage, taxes, leverage or individual time horizon.
Market structure can change. Molnar’s 2026 preprint discusses deterioration in several Bitcoin cycle indicators. It is preliminary research, not conclusive validation or refutation of this model.
Pi Cycle methodology and data sources
The calculation uses completed daily BTC/USD closes, including the selected day. The current UTC day is excluded. Missing dates reset the averaging window; prices are never filled forward.
Daily CSVJSON datasetMethod specification v1First-publication log
Exact calculation and reference pseudocode
- Sort positive, finite prices by their stored calendar date. A missing date restarts the consecutive-day window.
- Calculate the unweighted means of the last 111 and 350 daily closes, including the selected day.
- Multiply the 350-day mean by two, then divide the 111-day mean by that threshold.
- Record an upward crossing only when yesterday’s valid ratio is below 1.00 and today’s is at least 1.00.
# Requires 350 consecutive daily closes; slice end is exclusive
sma111 = mean(close[t-110 : t+1])
sma350 = mean(close[t-349 : t+1])
pi_ratio = sma111 / (2 * sma350)
signal = pi_ratio[t-1] < 1 and pi_ratio[t] >= 1A valid 111-day average appears after 111 consecutive observations; the ratio requires 350. Weekly and monthly displays use the last available daily observation in each period. Every calculation still uses daily data.
Source chain, closing times and revisions
January 4, 2011–March 13, 2026: historical import. The original provider and closing cutoff are not recorded in the available source metadata.
From March 14, 2026: CoinGecko ticks aggregated by BitcoinReversal. The close is the last available tick in each completed UTC day. Sampling or outages can differ from an exchange’s official daily candle.
Dates use the canonical stored price_date at 00:00 UTC. Different vendors, exchange composites and closing cutoffs can produce different values or crossing dates. Completed-day results can change if the source history is corrected; no future prices enter the moving averages.
The API refreshes its derived cache within five minutes when the source contains a new completed day. The visible observation date and freshness label identify the data actually available.
Versions, checksums and first-publication records
Method v1: the fixed 111-day / doubled 350-day specification. Calculation implementation: 8.0.1. Presentation: 15.0.0, reviewed September 8, 2026.
September 5, 2026 date correction: crossing labels moved one day forward when older local-midnight timestamps were replaced with the stored calendar dates. Underlying price observations and formula were unchanged.
The publication log starts with this release and records each newly available completed day once. It is not backfilled. Existing entries are retained if the source history is later corrected. This is a publisher-maintained record, not independent timestamp certification.
Each entry includes its first recorded time, exact reading, method SHA-256 and input-history SHA-256. The API identifies the method used. Report a data or methodology issue.
How Pi Cycle fits alongside other Bitcoin models
Opportunity Score
A 0–10 historical ranking of price structure. A separate original model; not a probability of future gains.
Bitcoin Power Law
Price relative to a fitted long-term growth trend. Extrapolation and fitting assumptions matter.
Stock-to-Flow
Price versus a scarcity-based reference. Supply scarcity alone cannot explain demand.
A dated reading you can reference
Credit the original indicator and this independent calculation. The citation below uses the latest completed observation, regardless of the historical date selected in the chart.
Bitcoin Pi Cycle Top questions
Has Pi Cycle identified every Bitcoin top?
No. The April 2021 crossing was followed by a higher nominal close in November without another crossing. The 2025 example above also shows that a low ratio can precede a substantial fall.
Does a ratio of 0.419 mean a 41.9% chance of a top?
No. The ratio divides one moving average by another threshold. It is not a calibrated probability, valuation score or forecast.
Can the signal repaint?
The current incomplete UTC day is excluded. A completed-day reading is stable for unchanged inputs, but historical source corrections can change past values. The first-publication log preserves the first recorded reading from its introduction onward.
Why does my chart provider show another crossing date?
Providers can use different exchange composites, closing times or date labels. This page calculates directly from its disclosed stored daily series. The daily CSV makes those inputs and signal dates inspectable.
Is Pi Cycle a standalone trading strategy?
No. It defines one rare trend-crossing condition. It does not define position size, entries, exits, execution costs or suitability for an individual investor.