The four-year cycle
Descriptive, oversold as prediction
Bitcoin is said to move in four-year cycles anchored to the halving: accumulation, a rally peaking roughly a year to eighteen months after the halving, then a deep bear market. The cycle is used to date tops and bottoms in advance.
- 2012 halving
- 2016 halving
- 2020 halving
- 2024 halving
Our own calculation on Coin Metrics Community Data (BTC reference rate, daily, through 14 August 2026). Each line starts at 100 on its halving day and runs 36 months. The 2024 line is incomplete because the cycle is not over.
How it is built
The halving dates are known in advance, so the model consists of counting months from each one and asserting that the same sequence repeats. The “lengthening cycles” variant adds that each cycle takes longer and returns less than the last.
Our verdict
Descriptively useful, oversold as prediction. Crypto does move in long waves; the halving is a real event. Neither fact licenses a calendar.
The record
Every line below is our own calculation on the public price series — reproducible where it is our own calculation, and linked to its primary source where it is quotation.
| When | What the chart says | What the data shows | Source |
|---|---|---|---|
| 2012-11 → 2016-07 | Each halving is followed by the same rally shape, peaking roughly twelve to eighteen months later. | The 2012 cycle peaked at month 12 at about 82 times the halving-day price; the 2016 cycle at month 17, at about 23 times. Two different months, and an amplitude smaller by a factor of three. | Moonkelp’s own analysis |
| 2020-05 | The pattern repeats. | The 2020 cycle peaked at month 18 at about 7.6 times the halving-day price — down by another factor of three. | Moonkelp’s own analysis |
| 2024-04-20 | The 2024 halving should have started the same sequence again. | Bitcoin’s all-time high for that cycle so far came in month 18: $124,824 on 6 October 2025, against $64,908 on halving day — a multiple of 1.9. The sequence 82×, 23×, 7.6×, 1.9× is a pattern fading out, not repeating. | Moonkelp’s own analysis |
| 2024-03-13 | The halving is what starts the rally. | Bitcoin set a new all-time high of $73,082 five weeks BEFORE the April 2024 halving — the first time the cycle high-water mark was passed ahead of the event that is supposed to cause it. | Moonkelp’s own analysis |
| 2025-12 | The cycle is the industry’s shared framework for what comes next. | Not any more. Grayscale told clients in December 2025 that it expects “rising valuations in 2026 and the end of the so-called ‘four-year cycle’”. By early 2026 the list of people publicly calling it dead included Cathie Wood, Arthur Hayes, Ki Young Ju, Raoul Pal and Bitwise’s Matt Hougan and Hunter Horsley — several of whom had built on it. | Ciaran Lyons, “The debate over Bitcoin’s four-year cycle is over: Benjamin Cowen”, Cointelegraph Magazine, 6 March 2026 · archived |
| 2026-03-06 | Once the calendar slips, the theory is abandoned. | It is reinterpreted instead. Benjamin Cowen, who has argued the cycle case longest, told Cointelegraph Magazine: “I honestly don’t know why people keep saying it’s not the thing. I mean, Bitcoin tops when it always tops, like every cycle, it tops in the fourth quarter.” His reading of 2026 is “a bear market year”, with a bottom around October 2026 — a claim still open at the time of writing, and one this page does not call either way. | Ciaran Lyons, “The debate over Bitcoin’s four-year cycle is over: Benjamin Cowen”, Cointelegraph Magazine, 6 March 2026 · archived |
Why the method does not hold
01
There have been very few completed cycles. Three or four observations of a repeating shape is an anecdote, not a statistical regularity, and it cannot distinguish a real cycle from three bull markets that happened to end.
02
When a cycle failed to arrive on schedule, the theory was amended rather than abandoned — cycles were declared to be lengthening, or stretching, or shifted. Each amendment saves the model and removes a little more of its ability to be wrong.
03
The mechanism weakens on its own terms: each halving cuts a smaller absolute amount of new supply, while demand-side shocks have grown. A model built on a supply effect must explain why that effect matters more as it shrinks.
The strongest case for it
The supply reduction is real, and reflexivity is real: if enough participants believe in the cycle, their behaviour can produce something that looks like one. Liquidity conditions also run in multi-year waves, and those genuinely move risk assets. The honest version of the claim is “crypto has long, violent boom-bust waves” — which is true and useful. The unsupported step is putting dates on them.
What to look at instead
Where a cycle model asserts a date, the descriptive gauges report a state: MVRV and the Mayer Multiple say how stretched price is right now, and the Fear & Greed index says how the crowd feels about it. None of them will tell you the month of the top, which is the point.
Questions
Is Bitcoin’s four-year cycle dead?
The pattern people mean by it has decayed to the point where it no longer constrains anything. Measured from each halving day to the highest price that followed, the multiples are roughly 82×, 23×, 7.6× and — so far in the current cycle — 1.9×. Long boom-and-bust waves are still real; a four-year clock you can set a date by is not.
Why did the four-year cycle stop working?
Its mechanism gets weaker every time. Each halving removes a smaller share of total supply, so the supply shock that the theory rests on shrinks with every repetition, while demand-side forces have grown much larger. A model whose engine fades cannot keep producing the same output.
What about “lengthening cycles”?
That is the amendment made when the schedule slipped: cycles were declared to be getting longer. It may even be true, but as an argument it is a rescue — each adjustment preserves the theory by removing another way for it to be wrong. Once a model can absorb any timing, it has stopped forecasting timing.
Sources & method
Every figure on this page comes from one public price series and arithmetic anyone can repeat. Where a claim would need a source we could not verify, it is not on this page.
- Coin Metrics Community Network Data — BTC reference rate and hash rate, daily — the price and hash-rate series every figure on this page is computed from
- Moonkelp’s own analysis — the calculation itself — method described above, reproducible from the series
- Ciaran Lyons, “The debate over Bitcoin’s four-year cycle is over: Benjamin Cowen”, Cointelegraph Magazine, 6 March 2026 · archived — Cowen’s defence of the cycle in his own words, and the named list of industry figures who by then considered it dead
The other charts
Perspectives, not investment advice. This page criticises a method, not the people who publish it — and it says so where a construction does something real. How Moonkelp works
This study is reviewed quarterly; corrections normally ship within days. Corrections:contact@moonkelp.com