Every bitcoin cycle produces a new explanation for why this time is different. In 2013 it was Silk Road and the Cyprus crisis; in 2017 the ICO wave; in 2021 institutions and pandemic money; in 2024 and 2025 the ETFs. The narratives change completely — and the curve keeps its shape: rise, overshoot, deep drawdown, a higher plateau than the cycle before, silence, repeat. The tempting conclusion is that the explanations explain nothing. The precise conclusion is more careful, and the care is where the insight lives: the narratives do not explain the shape of the curve — at most they decide when the machine runs. Without a demand wave the engine idles; in that sense the stories are not nothing, they are the fuel. But fuel does not determine the design of an engine. The overshoot, the rhythm of capitulation and mania — that comes from a mechanism indifferent to whichever story is being told. What follows is that machine, opened into its three layers — each answering one question none of the others can touch.
Current Conditions
Clearing the Ground
First, clear the ground. The classic price theories all fail the same way: stock-to-flow sold correlation as causation and was empirically dismantled in 2021–22; production-cost models read the causality backwards — mining costs follow price, not the reverse; network-effect models describe growth phases but explain no price point. All three pretend a valuation anchor exists. None does: bitcoin generates no cash flows, so no valuation gravity pulls price toward a fundamental value. What exists instead is a cycle machine — and, the central refinement of this piece, a dial that sets how strongly it amplifies.
Layer One: The Bottom Is a Capitulation Machine
A cycle bottom is not a price level; it is a state of ownership distribution, reached when two capitulations have run their course. The miners’ first: when price falls below the production costs of inefficient operators, they shut down; hashrate declines; the difficulty adjustment then lowers costs for everyone who remains and for new entrants with better hardware and cheaper power. By the end of a bear market the surviving miner population produces at structurally lower cost than the previous cycle’s, and distress selling dries up — a process so regular that hash-ribbon crossings have marked nearly every bottom. It is the cost-pressure physics of The Hoarder’s Paradox running to exhaustion. One honest qualification: with roughly 450 BTC of daily issuance against billions in daily volume, miner capitulation now works less through the quantity sold than through the signal it sends. In parallel runs the holders’ capitulation: the late buyers of the previous top, underwater and exhausted, sell into weakness to buyers who then do not move for years. On-chain, the coin base visibly ages; at the bottoms of 2015, 2018 and 2022, roughly half the supply sat at a loss. The bottom is the point where the selling is complete — no one is left who can still panic.
This state has a measurable signature: the realized price — the aggregate acquisition cost of all coins, reconstructable from the chain — and historically, bottoms have turned almost exactly there. One distinction keeps this consistent with everything above: the realized price is a distribution mark, not a valuation anchor. It marks where capitulation exhausts its material, not what bitcoin is worth; a reservoir of sellers running dry, not gravity. And one honesty about the toolkit: realized price, hash ribbons and loss-share metrics were calibrated on retail-dominated cycles. Under ETF and custody structures, on-chain legibility loses resolution — coins in custody move by fund logic, not holder psychology. The bottom’s beautiful measurability was itself a feature of the old buyer base; the machine may keep its logic while losing its gauges.
Layer Two: The Overshoot Engine — and Its Dial
Why does bitcoin overshoot instead of settling? Three structural properties, stacked. Perfectly inelastic supply: every commodity in history has had a supply response — price rises, production expands, price is dampened; every gold rush is supply called forth by price. The difficulty adjustment severs exactly this channel: a tenfold price brings tenfold hashrate and precisely zero additional coins, because difficulty re-targets every 2,016 blocks. The entire incentive that would become extra production in any other good becomes, in bitcoin, extra security — the unforgeable costliness of The Same Trade, Judged Twice, viewed from the supply side. It is the first good with a dead supply-side price feedback — no price in the world can call forth a single additional satoshi — so demand fluctuations cannot drain into quantity and go entirely into price. Volatility is the flip side of hardness. Reflexive demand: for a monetizing good, a rising price is itself the argument to buy — more trust, more attention, a fatter security budget — and a falling price the argument to sell; positive feedback structurally produces overshoot, never smooth convergence. Leverage: price discovery now runs predominantly through derivatives, perpetual futures move multiples of spot, and leverage amplifies both directions until the liquidation cascade clears the excess.
Those three explain why the system can overshoot. Whether it does depends on who is buying — because reflexivity is not a constant of the system but a property of the marginal buyer. It is a dial. Reflexive capital — the retail buyer who buys because it is rising, levered, inside the attention cascade — turns it up: every move generates the next. Non-reflexive capital — the allocator buying programmatically, by mandate, without momentum logic or a leverage pyramid — turns it down: the demand acts, but it does not feed back. The evidence is the most striking market phenomenon of recent years: ETFs and corporate treasuries absorbed on the order of two million coins — roughly a tenth of the supply — without a 2017-style ignition. That does not show demand is inert; without that bid the price would stand materially lower. It shows that demand quantity does not determine price amplitude — the old amplitude never came from quantity alone, but from quantity times reflexivity times leverage, and allocator demand is the type with the feedback loop switched off. It absorbs instead of igniting. The machine has not lost its demand; it has lost its amplifier. (One calibration, for honesty: even allocator flows chase performance on longer time constants — the dial turns down, never fully to zero.)
The shape of the curve is made by the feedback property of the marginal buyer — not by the story he tells himself. The narratives are interchangeable. The buyer type is not.
Layer Three: The Top Is Where Mechanics End
For the bottom there are hard marks. For the top there is no counterpart — no realized ceiling, no valuation cap, no capitulation from above. The top is where mechanics end and only minds remain. What moves is the anchor: in each era the market forms a collective feel for what counts as “normally expensive,” and that feel migrates with the world — money supply, real estate, wages. A hundred thousand dollars per coin was absurd in 2017, bold in 2021, self-evident in 2025; the number did not change, the anchor moved. The top forms when price runs too far ahead of the migrated anchor and the marginal new buyer — measuring against the feel of his own era — simply stays away. Nothing more is required: because reflexivity runs both directions, absence suffices; momentum tips, and “it’s rising” becomes “it’s falling.” The top needs no rush of sellers, only buyer exhaustion — and it reveals itself only in hindsight.
Since the dial and the anchor both explain softening cycles, their division of labor matters: the dial sets how much amplification is in the system at all — how much reflexive, levered capital rides the wave; the anchor sets where a running wave exhausts — how far price can outrun the era’s feel. The observed dampening has two additive sources: less amplifier in the system, and a shorter leash for whatever still swings. Two empirical supports carry this. Measured against the money supply instead of in dollars, the all-time highs move closer together — a substantial part of each nominal rise is anchor migration, not revaluation. And the bottom-to-top multiples have collapsed — one hundred x, twenty x, three to four x — exactly as habituation predicts. Together, dial and anchor make a falsifiable claim, which almost no top theory dares: continued dampening of money-supply-adjusted amplitudes. A future cycle that grows larger in adjusted multiples would put both mechanisms in explanatory trouble. That is a strength. The honest limit stays attached: “feels too high” is useless as a real-time signal — nobody measures the crowd’s anchor live; the asymmetry is real and probably permanent, because the bottom is compulsion and the top is exhaustion, and compulsion is more measurable.
Timing: Two Candidates, No Verdict
The three layers explain the shape. When the machine runs, they do not explain — and the model owes the same honesty it demands of the narratives. Two serious candidates, which four data points cannot separate. The halving as Schelling point: whether the four-year rhythm hangs causally on the supply shock or on the fact that everyone’s expectations synchronize on it is unresolved — the cycle may work because everyone believes it works, and a Schelling point everyone couples to is real for as long as the coupling holds. And the global liquidity cycle: the bottoms of 2015, 2018–19 and 2022 sat in or near global liquidity turns, and an asset with no cash-flow anchor, trading pure monetary premium, would plausibly be the most sensitive instrument to the availability of investable capital — but n=4 discipline applies here too; the coincidence is a named candidate, not a verdict. Both can be true at once: a Schelling point synchronizing expectations inside a liquidity regime feeding the waves. The model delegates timing to the two — and keeps for itself what it explains sovereignly: the shape.
What to Actually Take From This
This is the machine-room companion to What the Price Is Made Of — and the model refuses the two lazy positions (“the cycle is law” / “the cycle is dead”) with something falsifiable in between.
Stop grading narratives; locate the layer. Every cycle’s story is fuel — interchangeable by construction. The productive question is never “is the ETF story true” but “which layer is doing the work right now”: capitulation mechanics, overshoot structure, or anchor exhaustion. The machine explains the shape; the fuel only the start time.
Watch the dial, not the volume. Two million coins absorbed without ignition settles it: amplitude was never about how much is bought but about whether the buyer feeds back. Reflexive, levered capital makes 2017s; programmatic capital makes quiet floors. The single most informative market variable is the composition of the marginal bid — and it is measurable, unlike the crowd’s anchor.
Trust the machine, audit its gauges, respect its silence. The bottom’s marks are real but calibrated on a buyer base that is migrating; the top has no marks at all; timing belongs to two candidates n=4 cannot separate. Position for a damped oscillation — softer drawdowns, smaller multiples, duller volatility — and treat any claim of precision beyond that as someone selling you their story as your signal.
Instrument Check — Worth Your Attention
Study — realized price and supply-in-loss: build the bottom gauge yourself. The public chain contains the whole Layer-One toolkit: reconstruct realized price from last-moved values, plot share of supply at a loss, and mark the 2015, 2018 and 2022 bottoms. The Down to the Metal move again — an afternoon of work, and the capitulation machine stops being a claim. Then note how ETF custody blurs the same gauges going forward: that blur is Layer One aging in real time.
Read — Soros’s reflexivity essays, the theory behind the middle layer. The overshoot engine is applied Soros: prices that alter the fundamentals they supposedly reflect, feedback replacing equilibrium. Read the core essays and notice how precisely bitcoin’s monetization dynamics fit a framework written for currencies and credit — and where the difficulty adjustment makes bitcoin the purest reflexivity laboratory ever built: the supply valve is welded shut.
Follow — the mechanics under the machine: The Hoarder’s Paradox, The Same Trade, Judged Twice and What the Price Is Made Of. The cost-pressure logic that makes miners structural sellers and sets capitulation physics, the unforgeable costliness that makes the supply side incorruptible, and the factory-and-buyer analysis this piece’s dial extends. Together with this one: a complete stack, from what is priced to why it moves the way it does.
Flight Log — Dispatch From Altitude
Every aircraft carries a hidden oscillation, and every pilot has met it. Disturb an aircraft in pitch — a gust, a thermal, a careless input — and let go, and it does not simply return to level flight. It enters a long, stately cycle: the nose drops, speed builds, lift grows, the nose rises, speed bleeds, lift decays, the nose drops again — a slow exchange between airspeed and altitude with a period of half a minute or more. Aerodynamicists call it the phugoid, and its first property is the whole first half of this piece: the shape of the oscillation comes from the structure of the system, not from the character of the disturbance. Gust or thermal or ham-fisted input — the phugoid neither knows nor cares. The period and the pattern are set by the machine: by how energy is stored and exchanged between speed and height. The disturbance only decides when it starts. Fuel, not design. Any pilot who tried to understand the phugoid by analyzing the gust would be laughed out of the briefing room — yet that is precisely how most market commentary treats each cycle’s narrative.
The second property is the dial. Whether a phugoid grows, persists or dies depends less on the disturbance than on who is flying. Hands off, in a well-behaved aircraft, it is weakly damped — it swings for a long time. A pilot chasing the oscillation with aggressive, reactive inputs — pulling because the nose is low, pushing because it is high, always slightly late — feeds energy into exactly the cycle he is fighting: pilot-induced oscillation, the aeronautical name for reflexivity, every response amplifying what it responds to. And a flight-path-stable autopilot kills the phugoid almost entirely — not by being smarter about gusts but by responding to deviation programmatically, without momentum-chasing, on a control law instead of an emotion. Same aircraft, same gust, three completely different amplitudes — because the amplitude was never in the disturbance. It was in the feedback property of whoever holds the stick. Retail hands make pilot-induced oscillations; allocator autopilots make absorption; the aircraft, like the asset, only supplies the structure.
And aviation adds the long-run coda: transport aircraft are deliberately certified and flown so that the phugoid barely matters anymore — augmentation, autoflight, procedures; the oscillation is still in the physics, but decade by decade less of it reaches the flight path. Nobody calls that the death of aerodynamics. It is the same aerodynamics with the amplifier turned progressively down — a damped oscillation on its slow way to boring, which in a passenger aircraft is the entire point. Watch the bitcoin chart with a pilot’s eyes, then: the swing is the phugoid — structural, indifferent to the story of each gust; the amplitude is the stick — set by whether the hands on it chase momentum or follow a mandate; and the flattening, cycle over cycle, is not the machine dying. It is the autopilot engaging. The disturbances will keep coming, and each will bring its own weather briefing full of narratives. The aircraft has heard none of them. It just swings the way it is built to swing — a little less each time someone steadier takes the controls.