What if you could predict
Bitcoin's price —
years in advance?

The claim sounds absurd Everyone who predicted Bitcoin has been humiliated

• Stock-to-Flow promised $100k by 2021 — and broke.
• Wall Street targets have missed by orders of magnitude, in both directions.
• Standard finance says it's impossible by definition: prices are supposed to be random walks.

And yet, hiding in plain sight, there is a pattern — you just need to look at the chart the way a physicist would.

The reveal Same data, three ways of looking

① The chart everyone knows: a hockey stick. Fifteen years crushed into a corner. Unreadable.

② Make the price axis logarithmic — each gridline is ×10. Now you see waves… around something that keeps bending.

③ Make time logarithmic too — Bitcoin's age, in doublings. Watch the year ticks crowd to the right: time itself compresses. The bending stops. Sixteen years, six orders of magnitude — from cents to $100,000+ — one straight line.

Is it real? Fit it on the past — it keeps calling the future

P(t)  ∝  t^5.6   ·   R² = 0.96 (what that means: next slide)

A walk-forward test: the red line is fitted on 2010–2020 data only (navy); the orange years 2021–2026 are unseen by the fit, yet land within ~20% — through the 2021 mania, the 2022 collapse and the ETF era. Nor is this hindsight: public versions of this same line have existed since 2018 (Santostasi) and 2019 (the “power-law corridor”), and the price has kept landing on them. It is one law: the 2010–2020 fit (β = 5.68), the full-period fit (5.62) and the published 5.7 agree within the error bars. Even a skeptical 2026 statistics paper conceded it out-forecasts every fancier model at 1–2 year horizons.

How good is "one line"? R² = 0.96 — and why that number is the whole story

R² measures how much of the variation the line explains: 0 is noise, 1 is perfect. The bottom panel is the leftover — price divided by the line: every mania and every crash is a ×3 wiggle that returns to 1. The cycles everyone trades are the residual; the line is the signal.

Before going further We ran the same ruler over everything else

Maybe every asset does this, if you squint? We fitted the same power law to the others:

• Apple, or any Mag-7 stock — tracks earnings, punctuated by products and buybacks. No single line survives.
• The S&P 500 — roughly exponential (~7%/yr): a different curve family altogether.
• Gold — the right general shape, but noisy and convention-dependent. Nothing like R² = 0.96.
• Ethereum — the closest cousin, still far off: β ≈ 2.1 at R² = 0.86, and its exponent drifts ~5× more than Bitcoin's.

Sixteen years on one straight line is not a market regularity — it is an anomaly unique to Bitcoin. Whatever is doing this is doing it to exactly one asset on Earth. That is worth understanding.

The full comparison — Ethereum, Litecoin, gold, stocks, with the receipts — comes later in the deck.

One minute of math A straight line on a log-log chart is called a power law

price  =  A × age^β

• A is just the line's starting height — the price at age one. It slides the whole line up or down and says nothing about growth.
• β — the exponent — is the slope of the straight line: how much growth you get per doubling of age. All the structure lives here.
• For Bitcoin, β ≈ 5.6. Concretely: every time Bitcoin's age doubles, its price multiplies by about 50 (2^5.6).

Age 2 → age 4: $1 → $50. Age 4 → age 8: $50 → $2,500. Age 8 → 16: $2,500 → $120,000. That's the last sixteen years, in one sentence.

Why this matters Power laws are how nature signs its work

No central planner, no schedule — self-organizing systems with feedback, everywhere you look.

And the two highlighted cards are the entire recipe — hold that thought.

Where we're going Two of these lines, up close: Bitcoin & Pareto

Power laws come in two kinds. On the left, a law of time — Bitcoin's price climbing at slope +5.6: the rare kind for a financial asset. On the right, a law of distribution — who has how much: human wealth, the Pareto law (1896, the “80/20 rule”): climb toward the richest and wealth rises as rank^1/α (slope +0.83), anchored on the 59-millionth-richest person at ≈ $1M. Two straight lines, two unrelated worlds. Keep both in view. By the end, they'll turn out to be the same story — and a surprisingly simple one.

The rare kind Finance is full of power laws — but not this kind

• Power laws of distribution are everywhere in finance and economics: wealth (Pareto — the one we just saw), incomes, firm sizes, market caps, insurance losses, the fat tails of daily returns and crashes (Mandelbrot). Finance runs on them.
• A power law of price in time is another matter entirely. Standard finance says it should not exist: a price is supposed to be noise around a drift — not a sixteen-year curve you can draw with a ruler.

Among liquid financial assets, exactly one has a price power law in time: Bitcoin. Gold has the right general shape but data too noisy to qualify; stock indices grow roughly exponentially — a different curve family. One asset on Earth. That is what makes it extraordinary — and worth explaining.

Credit where due This line was discovered long ago

Physicist Giovanni Santostasi spotted Bitcoin's power law years ago (he published it on Reddit in 2018, from data he had studied since ~2014), and spent the following years refining it — culminating in a 2026 paper with Stephen Perrenod: β ≈ 5.7, R² = 0.96, plus a first proposal for decomposing the exponent.

He also showed the price isn't alone: hash rate, addresses and price are all power laws of each other — a whole self-consistent system.

He tells the whole story in his book, The Physics of Bitcoin (2026) — “not an asset but a force of nature”.

A note on 5.7 vs 5.6: same law, different cutoff. They report ≈5.69; our refit through mid-2026 gives 5.62 ± 0.21 — identical within honest error bars (the exponent also depends on where you start the clock). We use ≈5.6 throughout; nothing hinges on the third digit.

But one question stayed open — the only one that really matters:

why ≈ 5.6?

Galileo vs Newton Measuring a law is not explaining it

Step one Splitting the mystery in two FRONTIER · APRIL 2026

This is brand-new science: Santostasi & Perrenod's decomposition was published in April 2026 — the field is moving in real time. Their whole thesis in one line: two power laws chained together, each doing a different job:

“Decomposing the exponent” just means: a power of a power multiplies. If holders grow as t³ and price grows as holders^1.84, then price grows as t^{3 × 1.84} = t^5.5.

β  =  β_A × β_M   →   3 × 1.84  ≈  5.5  (measured directly: 5.6 — the ~2% gap is regression noise, well inside the error bars)
one mystery number → two smaller ones, each needing its own explanation

The itch Something bothered us: the network is counted twice

• The 3 is network spreading — the idea propagates from person to person.
• The 1.84 was labeled “Metcalfe” — network value, worth more when more people connect.

The same word, network, doing the work in both factors. If the network is what spreads the idea and what prices it, the explanation is circular — a red flag, not an answer. That itch is what started the whole investigation. The way out: follow what actually goes through a new adopter's head — it turns out to be two decisions, not one.

Two decisions, two mechanisms. Metcalfe explains the first (adopt), wealth explains the second (how much). Putting the network in both factors counts it twice — once for spreading the idea, again for pricing it. Once Metcalfe is correctly parked inside the 3, β_M is free to be what it really is: the size of the cheque each new believer writes.

Monetary theory agrees (Menger; Ammous): a money is first a store of value, then a medium of exchange, then a unit of account. Metcalfe prices the exchange stage; Bitcoin is winning the store-of-value stage — which is decision ②, an allocation of savings.

First factor — the lived experience Bitcoin spreads like a virus — we all lived it

Bitcoin is caught by understanding, not by handshake — but the stages are an epidemic's, and every Bitcoiner has lived them:

Organizations go through the same stages — with committees: slower, but carrying far more capital.

The textbook case: public dismissal → conversion → becoming a super-spreader. Every bitcoiner recognizes the stages. The real question: is the pattern quantitative — does an idea-epidemic have a law?

First factor — the theory The 3 already has one — epidemics

Santostasi & Perrenod's explanation of the first factor: Bitcoin adoption behaves like an epidemic, and epidemic mathematics fixes its pace.

The result they build on — you will hear the name again — is Colgate (1989): Stirling Colgate, an astrophysicist modeling the early AIDS epidemic, showed that on real human networks — tight communities, loosely bridged — an epidemic does not explode exponentially. It has to grind its way from group to group, and the headcount grows as the cube of time, t³. Forty years old, replicated across diseases and networks since.

Their claim, then: Bitcoin adoption is such an epidemic, so its headcount should grow as t³. But why would a mental virus obey the same math as a biological one?

First factor — the topology Why a contagion of the mind obeys the same math

What sets the pace of an epidemic is the shape of the network it travels on, and beliefs travel on the same clustered human network as diseases: families, workplaces, cities, timelines.

• Tight clusters, loose bridges. We mix densely inside groups, rarely between them. A contagion saturates one cluster, then waits for a rare bridge to the next — advancing layer by layer, which bends exponential into polynomial (Colgate's t³). It's the topology, not the biology.
• But there's a catch — and it makes AIDS the right parallel, not the flu. An airborne virus spreads whether you decide anything or not — you just breathe — so in a mixed crowd it goes exponential (COVID only turned polynomial, t^2.1, once lockdowns depleted the susceptible pool*).
• HIV and Bitcoin both spread by a decision. HIV needs a repeated behavioural choice; Bitcoin needs a cognitive one. That choice is what confines spread to stratified sub-communities — so the polynomial law is intrinsic, no lockdown required. Same reason each carries a t³.
• And we measured the boundary. Mobile phones — the biggest tech adoption ever, but with no worldview gate and carriers pushing supply — grew exponentially (×1.6/yr for 25 years, then saturation; our Test 79). No gate, no t³. The choice is what makes the polynomial.

*COVID subexponential t^2.1±0.3 via containment: Maier & Brockmann, Science 2020. Intrinsic polynomial spreading on clustered human networks: Colgate 1989 (AIDS, t³); Szendrői & Csányi 2004; Chowell 2016 (review). Communication: Leskovec–Adamic–Huberman 2007 (~linear cascades); Goel–Anderson–Watts 2016 (1B Twitter cascades); Bass 1969. Marketing — seeding, social proof, astroturfing — is applied epidemiology of belief.

First factor — the payload Why people catch it — and why the answer doesn't matter

Different layers catch different strains — necessity in weak currencies, elegance for engineers, scarcity for investors. And here is why the argument doesn't matter: the theory does not require these beliefs to be true — only contagious. The math runs on transmission, not on truth.

First factor — who catches it first Who catches it first isn't random — it's psychology

Adoption of anything new follows the same curve — innovators, early adopters, majority, laggards (Rogers, 1962). What decides where you fall is susceptibility to a worldview change, and Bitcoin demands a big one: that money needn't be issued by a state. (Strictly, catching it is exposure × susceptibility: you can be perfectly susceptible and never have truly met the idea — or hear about it every day and stay immune. The epidemic needs both to line up; psychology governs the second.)

• Individuals — analytical, contrarian, low-trust-in-institutions minds first; the socially-anchored last. (Anecdotal but striking: bitcoiners skew ~50% INTJ/INTP, ~13× over-represented.)
• The wealthy — susceptibility is independent of net worth: a contrarian billionaire (Draper, Salinas) adopts early, a cautious one late. Wealth sets how much, not when.
• Companies & funds — committees, fiduciary duty and career risk slow them down; the 2024 ETF was the threshold that flipped social proof from barrier to accelerant.
• Banks & central banks — latest, and rationally: a conflict of interest — they are the incumbents Bitcoin routes around. Late because threatened, not uninformed.
• Countries — two channels (Tests 59–60): innovation (US, Singapore first) + monetary necessity (Turkey, Argentina). Stable-currency Switzerland under-adopts — no pain, no urgency.

Across every stratum, susceptibility and wealth are independent axes. Psychology decides when you catch it; wealth decides how much you can bring. Keep that in your pocket — it becomes the key to the second number.

*Rogers diffusion; Granovetter threshold model; personality-cohort survey data. Illustrative, not load-bearing — the formula stands on the wealth exponent alone.

First factor — see for yourself A real virus and a mind-virus — the same slope

Left: cumulative US AIDS cases — Colgate's own subject — on four CDC-sourced points (1981–1988), a near-perfect straight line on log-log (R² = 0.995), growing as ≈ t³ — not exponentially. Right: Bitcoin's adopter count on the same axes — a first glimpse of the test to come. A biological contagion and a cognitive one, spreading on the same human network, obeying the same power law of time. The exponent shifts a little with the assumed epidemic start (~2.7–3.0) — the same time-origin sensitivity we're honest about for Bitcoin.

First factor — another mind-epidemic Domain names: the internet itself spread as t³

.com/.net registrations (Verisign) are a pure cognitive epidemic — no biology at all, just the idea of being online. Its adopter base grew as t³ (β_A ≈ 3), on a completely different network, a decade and a half before Bitcoin existed. A physical virus, then a mind-virus — same number.

First factor — the receipts Not an analogy — four decades of independent science

Biology, marketing, social media, pure math — every one of these lands sub-exponential: polynomial, never explosive. And the epidemics that run society-wide and can be measured end-to-end — AIDS (≈ t³), internet domains (≈ t³), the broadest collaboration networks (up to t^3.1) — keep landing at ≈ 3: a number intrinsic to the human network, physical or mind. Which sets up a hard prediction: if Bitcoin really is a society-wide cognitive epidemic, its adoption curve has no choice. It must read ≈ t³.

First factor — the moment of truth So Bitcoin has no choice: it must read t³. Does it?

Three observables that share no data source: on-chain addresses holding bitcoin, cumulative views of the “Bitcoin” Wikipedia article, and cumulative worldwide Google searches for “bitcoin”. Awareness, curiosity, and action — all growing as ≈ t³ (slopes 3.0–3.6): the cube Colgate's mathematics predict.*   *and reproduced by our network simulations.

Bam — 3. Three rulers, no shared data, one exponent: the number four decades of epidemiology said it had to be.

First factor — nailed Case closed — the 3 belongs to the network

• Theory: forty years of epidemiology and network science predict polynomial, ≈cubic spreading on human networks (Colgate 1989, and a large literature since).*
• Other networks: the same exponent wherever ideas spread — AIDS ≈ t³ (a real virus), .com domains ≈ t³ (15 years before Bitcoin), academic collaboration networks t^2.3–3.1. Not a Bitcoin quirk.
• Bitcoin data: three independent observables give 3.0–3.6 — the predicted number, measured.

“Boring” but quietly profound: the 3 is really a measurement of the human social network itself — the clustered, scale-free shape of how we connect. Bitcoin's adoption curve is a ruler laid against the topology of humanity. Any spreading idea would read the same ruler — including the famous network effect: “I join because others are already there” is Metcalfe's force, and it is exactly what spreads the idea. Metcalfe lives here, inside the 3.

And so does the clock. The 3 now owns two things: the network — Metcalfe included — and all of the time-dependence. Whatever the 1.84 is, it cannot be the network again, and it cannot be time. Bitcoin has something special — and it lives in that 1.84.

*Szendrői & Csányi 2004, Chowell et al. 2016, Williams & Chen 2025, among others.

Second factor The mystery: 1.84

When the number of holders doubles, the price doesn't double — it multiplies by 3.6 (2^1.84).

Each new wave of adopters somehow moves the price more than proportionally. Who ordered the extra?

Notice what the 1.84 is not: there is no clock in it. A doubling of holders took months in 2011 and takes years today — yet each doubling multiplies the price by the same ×3.6. All of the time lives in the 3; the 1.84 is pure leverage per doubling. And for 3 × 1.84 to compose into one clean power law of time across six orders of magnitude, that leverage must be the same at every scale — the signature of another power law: not one in time, but a distribution. Which one?

Before the right answer — the wrong ones. We know they're wrong because we tried them ourselves.

*Holders = blockchain addresses holding a balance (~57M today); 1.84 is the fitted slope of price vs holders, both in log, over 2010–2026. It is measured on its own — not obtained by dividing 5.6 by 3 (which would give 1.87; the ~1.5% difference is fit noise between the three separate regressions). And yes: in expanding windows the two fitted components drift in opposite directions — we chased that (Tests 21/35/73) and it turned out to be regression mechanics plus the address-proxy aging, not physics: the product stays put, and the price-vs-holders line itself stays straight. The straight line is the claim.

Second factor Three seductive wrong answers

• “It's Metcalfe — network value.” Metcalfe's law is solid science in its home domain: a communication network's value to its members grows as N², the number of possible connections. It priced telecoms and social networks well. But we buried this one ourselves a few slides ago: connection-value is why the idea spreads — it already lives inside the 3. Making it the second factor too counts the network twice (the itch from slide 10). And the data agrees: 2.0 ≠ 1.84, and in 2009–2010 the network grew while the price stayed at zero.
• “It's money printing.” We spent 20+ tests on M2, inflation, Cantillon effects. The effect is real — and cancels out of the exponent almost exactly.*
• “It's three effects added up.” Our own first paper split 1.84 into wealth + monetary + growth components. The algebra worked; the causality didn't survive our own re-testing.

*In total: 79 tests, 20 simulations, and 24 documented dead ends — our full false-leads register is public.

Second factor The clue was on-chain all along

Whatever explains 1.84 must explain this: the capital invested per holder has grown ×2,500 in 14 years — a clean power law of the number of holders. If adopters were arriving in random order, this line would be flat. Each wave of adoption arrives richer than the last.

Second factor Bingo: the oldest power law in economics

Remember the second line we told you to hold onto? Vilfredo Pareto, 1896: human wealth follows a power law. The “80/20 rule” is its folk version. Its steepness is a single number, α — measured by economists for 130 years, on tax records and wealth reports, without ever looking at Bitcoin:

α ≈ 1.2

Bitcoin's idea-epidemic spreads through that pyramid: geeks → entrepreneurs → funds → BlackRock → states. Each wave taps proportionally deeper into it. And only 21 million coins exist — every richer wave must buy them from earlier holders, at whatever price releases them.

Second factor — the result 1.84 is wealth inequality itself

β_M  =  1 + 1/α  =  1 + 1/1.2  ≈  1.82

• The “1”: in a perfectly equal world, every adopter brings the same capital — price simply tracks adoption.
• The “1/α”: in our unequal world, later waves are richer — an inequality premium of +82%, set entirely by the shape of the global wealth pyramid.

And the circularity is gone: the 3 lives in the network of minds (how fast the idea spreads); the 1.84 lives in the distribution of wallets (how much capital each wave carries). Two different objects — and the second is measured entirely outside Bitcoin.

Second factor The wave, with real names

Illustrative public milestones; sizes are the order of magnitude of wealth commanded by each new actor class (not capital allocated). Sixteen years, eight orders of magnitude: each wave reaches actors commanding ~10–100× more capital than the last — the geometric climb the formula needs. The order of arrival is the susceptibility story from the first factor; the size of each wave is the wealth pyramid.

Which means, today Early or late? It depends which layer you sit in

While the first companies were putting bitcoin on their balance sheet, the individual wave was already ~7% in. Every heavier layer is still in its first movers — central banks hold 0% directly, though a few already own it indirectly (the Swiss National Bank holds MicroStrategy shares). And each layer buys its coins from the layers that moved earlier, at the price that makes them sell.

Confirmation 1 of 2 The number lands exactly where the pyramid says

(a) What we measured, and how: plot Bitcoin's price against its number of holders — each dot is one day, 16 years — and fit a straight line. Its slope is βM = 1.83: every doubling of holders multiplies the price by 2^1.83 ≈ 3.6. (b) What it means: that slope (with honest ±0.09 error bars) lands in the green band that global wealth inequality predicts — computed with no Bitcoin data at all. Zipf is excluded; Metcalfe (2.0) is 1.9σ off and has no mechanism.

Confirmation 2 of 2 The theory predicts who does NOT get the curve

A real theory needs a control group. The premium requires BOTH ingredients — absolute scarcity and an epidemic still climbing the wealth pyramid. Ethereum lacks the cap; Litecoin's epidemic died among retail. Both premiums: zero. Bitcoin's: +0.84, the size the pyramid predicts. (And Litecoin has survived since 2011 — so mere survival creates no curve.)

The complete picture Why not Apple? Or gold, or Ethereum? One table

The formula needs three things — a cognitive epidemic that climbs the wealth tiers, a fixed supply, and the wealth premium that follows. Line everything up:

Bitcoin is the only row that ticks every box. Ethereum has the epidemic but no cap; Litecoin has the cap but its epidemic never climbed past retail; everything else lacks a fixed supply entirely. The theory isn't meant to fit Apple — and that it doesn't is the point.

Bonus — a live stress test 2024: the ETF shock

One BlackRock address = millions of owners: the spot ETFs broke the visible headcount. The machinery reshuffled — from “more holders” to “same visible holders, double the capital each” — and the price stayed on the line. To be precise, this doesn't confirm the formula; it shows the law is structural, not an artifact of one growth channel — and the hand-over from headcount to capital-per-holder is exactly what a wealth-climbing wave does when it moves up a layer.

One more thing We did our best to kill our own theory

• Honest error bars: sixteen years of daily data is only ~4 independent market cycles — so we quote β = 5.62 ± 0.21, not the naive ± 0.02.
• Convention checks: the exponent depends on when you start the clock — so we verified the theory on the quantity that doesn't (the price–adoption relation).
• Mechanism stress tests: simulations with noisy adoption ordering, alternative orderings, alternative explanations — 79 tests, 20 simulations in total.
• Two of our own favorite charts died in the process — killed by our re-examination, removed, and logged in a public register of 24 dead ends.

The formula survived every attack we could invent — and came out sharper. That is what we mean by “demonstrated”.

The keystone A theory is born

β  =  β_A × (1 + 1/α)

epidemic speed × wealth inequality  =  3.0 × 1.82 ≈ 5.5  ·  measured 5.62 ± 0.21

Bitcoin's price curve is the speed of an idea-epidemic, multiplied by the inequality of human wealth.

Nothing left to tune. One input is external to Bitcoin entirely — α ≈ 1.2, the wealth Pareto exponent from a century of economics. The other, βA ≈ 3, is measured three independent ways on Bitcoin's own adoption data (addresses, searches, Wikipedia) — and it lands where spreading on human networks lands (AIDS ~t³, internet domains ~t³). Once those two numbers are in, the formula has no freedom left: it had to give ≈ 5.5, and the market says 5.62 ± 0.21. It could have missed — so far it hasn't.

📄 Full paper & all 79 tests: e-xtend.com/btc_power_law

How to read it Bitcoin has weather, seasons, a climate — and ice ages

Most market commentary is weather forecasting. Most “models” fit the seasons. This work is about the climate.

So — can you predict the price? What the law gives you, honestly

• The trajectory, not the drama: the law is the trend; the bubbles and crashes are oscillations around it, carrying no long-run information. Years-ahead prediction of the corridor — not of next month. Climate, not weather.
• Conditional, and monitorable: it holds while the epidemic keeps climbing wealth tiers and wealth stays Pareto-shaped. Both are observable dashboards, and both would bend the curve visibly before breaking it.
• Built-in calm-down: a power law in age means every doubling takes longer than the last. The insane early returns were young-epidemic math, not a promise.
• A horizon, not a cliff: saturation happens layer by layer, not all at once. Simple arithmetic on the t³ law: the individual wave (~7% in) wouldn't approach half-penetration before the late 2030s — and the corporate, institutional and sovereign layers have barely produced their first movers, extending the runway well beyond. Sixteen years of adoption data show no bending yet; when the hand-over to mature-asset growth comes, it will be gradual and visible years ahead.

Research summary — not investment advice.

The question everyone asks So… when does the line cross $1 million?

Nothing new is fitted here — it is the same line, extended: $1M around 2034 (2033–2036), $10M around 2047 (2045–2053). The ranges cover the honest error bars and the defensible choices of when to start the clock — note how the fan opens with distance. Read it as the corridor's centerline: cycles will overshoot and undershoot it, and the whole projection is conditional on the epidemic continuing to climb the layers (the runway slide). Watch it live: charts.bitbo.io/long-term-power-law.

Standing on shoulders Credits

• Giovanni Santostasi & Stephen Perrenod — the law, its measurement, and the first decomposition β = β_A × β_M. The Galileo work this builds on.
• Stirling Colgate (1989) — cubic epidemic growth on human networks.
• Vilfredo Pareto (1896) → Piketty, Davies, Credit Suisse — the wealth power law, α ≈ 1.2.
• This work — β_M = 1 + 1/α: the inequality premium, the wealth-climbing mechanism, the cross-asset control test, and the open test suite (79 tests, 20 simulations, 24 documented dead ends).

wineLightning · 2026 · draft v2 — all figures regenerable from raw data with one script.