← Daybreak
BITCOIN'S CYCLE · EXTRAPOLATED & STRESS-TESTED

The internet is full of Bitcoin cycle predictions. We built every model we could — then checked whether any of them can be trusted.

0 models·0-yr horizon

built on 0 completed cycles of evidence.

We extrapolate Bitcoin's four-year rhythm a decade out — to roughly 2036 — under one assumption: that the cycle repeats. The seductive price path is the hook. The honest part is what a held-back test against history does to it. It comes down to one question: does the boom keep shrinking?

A thought experiment about a pattern, not a price target. Built from eight models on Bitcoin's full daily history (2009 → mid-2026). We sell no predictions — the job is to separate what the data supports from wishful thinking.

PART 1 — THE CYCLE, MEASURED

Bitcoin booms on a four-year beat. Each boom is smaller than the last.

Before extrapolating anything, measure what's actually there. Bitcoin has run four full boom-and-bust cycles and is partway through a fifth. Around its long-run trend — a smooth path that has tracked sixteen years of price across a thousand-fold climb — price swings far above at each top and back below at each bottom.

Two things jump out. The rhythm's length barely moves: the mature cycles run about four years top to top. But the height of each boom keeps collapsing. It is not the beat that's changing — it's how high each one reaches.

cycle_anatomy

loading cycle_anatomy
Five tops, from $30 (2011) to the record $124,724 (Oct 2025). Each top sits a smaller multiple above the prior bottom than the one before.
THE BOOM IS SHRINKING — GROWTH FROM BOTTOM TO TOP, EACH CYCLE
0×2011 → 2013 boom
0×2015 → 2017 boom
0×2018 → 2021 boom
0×2022 → 2025 boom

cycle_period_amplitude

loading cycle_period_amplitude
Top panel: tops stay ~1,440 days (≈ 4 years) apart. Bottom panel: the boom's growth fades 604× → 107× → 21× → 8×. The beat holds; the height doesn't.
PART 2 — HOW THE CYCLE EVOLVES

Don't forecast the price. Forecast the pattern.

Here is the centerpiece. Instead of guessing a price, model how the cycle itself is changing — then carry that forward. Two regularities do the work: each boom comes in at about two-thirds the size of the one before (a clean, steady fade), and the tops stay near four years apart.

Run both forward and a price path appears. The next top (~2029) lands near $288K, the one after (~2033) near $540K, and the path reaches roughly $860K by the end of 2036. Seductive — but read the next panel before you believe it.

meta_model_projection

loading meta_model_projection
A fading boom on a rising trend. Projected tops ~$288K (2029) and ~$540K (2033); end-2036 near $860K. The shaded band is a faster- vs slower-fade envelope, not a confidence range.
PART 3 — THE MODEL MENU

Eight ways to run the cycle forward. They mostly agree — except on one thing.

One model is a single opinion. So we ran the cycle forward eight different ways — a hand-fit fading wave, a popular off-the-shelf forecaster, a self-fading wave, a flexible recurring rhythm, a bubble-timing model, a deep neural net, and a deliberate strawman. Each makes a different bet about how the future rhymes with the past.

What's striking is where they split. Almost all of their disagreement, ten years out, comes down to the one question from Part 2: does the boom keep shrinking, or not?

model_damped_cycle

loading model_damped_cycle
One four-year wave whose height keeps shrinking. Against the real cycles, its fitted tops land within days of the real ones. Forward: ~$370K (2029), ~$881K (2033), ~$1.7M (end-2036).

model_prophet

loading model_prophet
The popular tool, and the foil. Its four-year boom is fixed-size — it literally can't shrink — so it repeats the same boom forever (~$503K by 2036) and its honest range balloons past anything meaningful: a sign it's out of its depth this far out.

model_structural_ts

loading model_structural_ts
Lets the wave fade on its own estimate. The wave nearly dies within a single cycle; a rising baseline carries the path to ~$850K by 2036.

model_gp_periodic

loading model_gp_periodic
Assumes a recurring rhythm but grows less sure far out — its range fans about 9× wider by 2036 ($226K–$2.6M, middle ~$765K). Honest uncertainty, not false precision.

model_lppls_forward

loading model_lppls_forward
A bubble-timing model. It recovers the real Oct-2025 top to within a week, then points to a next top ~2029 near $167K — and honestly goes silent after that. It's built to call one peak at a time, not a decade.

model_nbeats

loading model_nbeats
A deep neural net, rolled a decade out. Its "cycle" flattens into a straight trend line (~$454K by 2036) — an honest demonstration that this kind of AI can't extrapolate a repeating cycle that far.

model_naive_repeat

loading model_naive_repeat
The strawman: the same wave with the shrink switched OFF. Freeze the booms at today's size and Bitcoin tops near $400K (2029), $1.0M (2033) and $1.95M (end-2036) — well above every fading-boom model.

models_converge_2036

loading models_converge_2036
Every path on one chart. Most models that let the boom fade cluster between about $450K and $860K by 2036 — though the slowest-fading wave runs to ~$1.7M, close to the no-fade assumption's ~$1.95M. Eight models, and the disagreement is almost entirely about how fast the boom fades.
PART 4 — THE RECKONING

Has guessing the next top ever actually worked?

A forecast that can't be tested is just a story. So we replayed history honestly: stand at each past top, use only the data available then, predict the next top — and compare it to what really happened. Both methods get a turn: "the boom repeats at full size," and "the boom keeps shrinking."

The verdict is uncomfortable for anyone selling a price target.

cycle_backtest

loading cycle_backtest
From the 2017 top, "boom repeats" predicted ~$308K for 2021 (actual $67.6K) — a 4.6× overshoot; "boom shrinks" said ~$87K (1.3×). From the 2021 top, "repeats" predicted ~$469K for 2025 (actual $124.7K) — 3.8×; "shrinks" said ~$182K (1.5×).
THE SCOREBOARD — HOW FAR EACH METHOD OVERSHOT THE NEXT REAL TOP
0.0×"Boom repeats" — average overshoot across the tests
~0.0×"Boom keeps shrinking" — average, far closer
0real tests on past cycles where the methods could even differ
0completed cycles in all of Bitcoin's history
PART 5 — PUTTING IT TOGETHER

A pattern worth watching. Not a price you can trust.

This is not advice to buy or sell, and it is not a price target. It's a map of what the cycle data supports as a scenario — and, just as important, how little of it survives an honest stress test.

What survives

If the four-year rhythm and its steady fade both continue, the next cycle tops somewhere in the mid-six-figures — the fading models cluster there. That's a scenario, not a target.

The one bet

Almost the entire forecast reduces to one question: how fast does the boom fade? Most fading models land between about $450K and $860K by 2036, though the slowest-fading wave runs to ~$1.7M; freeze the booms entirely and it reaches ~$1.95M.

The reckoning

Replayed against history, 'the boom repeats' overshot the next top about four times, every time. The shrinking version came far closer — but on only two real tests.

The honest size

Mid-six-figures next cycle, fading thereafter — held loosely. This is a thought experiment about a pattern, not a price prediction, and certainly not advice.

LIMITATIONS — READ THESE
  • Bitcoin has only ever completed three or four cycles. You cannot validate a ten-year cycle forecast on that — there is almost nothing to test it against.
  • Every model here assumes the past rhythm continues. Regular patterns break; a trend that has held for sixteen years carries no guarantee for the next ten.
  • "The cycle repeats" is the single most over-confident claim in crypto. The backtest shows why: the repeat-at-full-size assumption overshot the next top every time.
  • The price paths are extrapolations, not validated forecasts. The shaded bands describe faster-vs-slower fades, not the true odds of any outcome.
  • Descriptive, not predictive. These models describe a pattern in what already happened; they don't tell you what Bitcoin will do next, and nothing here is financial advice.

Descriptive research, not financial advice. Bitcoin is volatile and can lose most of its value in a year — as it has four times in this dataset. A repeating cycle is an assumption, not a fact.

Appendix — all 12 charts & model inventory →