No question in Bitcoin research is asked more often, or answered less consistently. Put the question to a room of analysts today and the answers will span capitulation, a bottoming process, and the early transition into the next expansion. The disagreement is rarely about the data. It is about definition. Cycle phases are still drawn largely by eye, and an eye tends to find what it went looking for.
Our research team has therefore designed a systematic regime detection framework built on price models extracted from the ledger, which makes it verifiable and behavioural rather than a matter of subjective judgement. It tracks the cost basis, meaning the average price actually paid, for three distinctive cohorts of investors who accumulated their coins over different time frames. The relative position of those three models tells us who paid the most and who the least, and that dynamic ordering constitutes the regime.
These are the three cohorts the framework is built on, ordered from the fastest to react to the slowest:
Set these three models against spot and against one another, and the ranking resolves into six possible orders. Each one is a regime.
The first test sets the side. When the newest money is paying more than the True Market Mean, the market sits on the bull side, which holds Breakout, Advance and Pressure. When it is paying less, the market is on the bear side of Breakdown, Capitulation and Repair. This is the depth of the market's valuation, and it is the most consequential single reading the framework produces.
The second test sorts within that side by asking where the recent cohort has settled, which is the spread. Take the bear side, where the market sits today. The newest money is below the True Market Mean in all three of these regimes, so it cannot be what tells them apart. The Short-Term Holder basis does. Above the True Market Mean, the market is in Breakdown. Below the True Market Mean but still above the newest money, Capitulation. Below the newest money as well, Repair, the newest buyers having begun to recover first.
Advance and Capitulation sit at opposite ends and are exact mirrors. In one the newest money pays the highest price in the market, in the other the lowest.
Table 1 puts that framework to work, recording how much of the past decade the market has spent in each of the six regimes. Every comparison carries a 5% deadband, so a regime changes on a decisive crossing rather than on noise. The measurement window opens at the December 2017 all-time high rather than at the start of the record. Market structure before that point was set without ETFs, corporate treasuries or institutional custody, and those older distributions travel poorly into the present.
The section above established the logic that defines a regime. Run that logic across every day of the record and the framework reads the market in two ways.
Figure 3 puts that plane on screen three times over, once for each cycle measured top to top. The first covers the 1,424 days from the 2017 all-time high to the 2021 one, the second the 1,427 days from there to the 2025 high, and the third is the cycle we are in now, which has not yet made a new high. Spread runs along the horizontal axis and Depth up the vertical, so every day sits somewhere on the disc as a single point, and the centre is where all three models agree.
Three lines cross there, one for each pairwise comparison. The horizontal is Depth = 0, where the newest money pays exactly the network average, and it separates the bull half above from the bear half below. The vertical is Spread = 0, where the Short-Term Holder basis sits on that same average. The third runs at 45 degrees, the parity line where Depth = Spread, which is the market in which the newest money and the Short-Term Holders have paid the same price as each other. Together they cut the disc into the same six wedges set out earlier.
Four things come off a dial. The wedge a point falls in names the regime. The order a cycle visits those wedges in is its rhythm, and because it turns rather than wanders, the direction of travel is a fair guide to which regime comes next. Distance from the centre measures intensity rather than identity, so a day deep inside the Capitulation wedge is further dislocated than one barely over the line. And the angle still to run before the next dividing line is how much of the current regime is left.
Today the market sits at depth — and spread —, inside the Capitulation wedge, where it has been for — days. Reaching Repair would take — percentage points of ground for the 1d–1w basis on the Short-Term Holder basis, and returning to the bull half would take — on the True Market Mean. Neither is near.
The first thing a regime tag on every day buys, as set out earlier, is history matching. Once today carries a label, it can be set beside every past day that carried the same one, and the price action that tended to follow can be modelled rather than guessed at. The dial has put the market in Capitulation for — days, so that is the comparison now worth making.
Figure 4 makes it. It shows the relative price performance of the last four capitulation regimes, the three completed ones and the one running now, each rebased to the day it began so the paths lie over one another. It reaches back to 2014 rather than to the 2017 window used elsewhere in the report, deliberately, because three completed episodes is a thin sample already.
The last three completed capitulations ran from September 2014 to July 2015, September 2018 to April 2019, and June 2022 to January 2023. They lasted 283, 205 and 208 days, averaging 232. At — days this episode has covered roughly a fifth of that, which is context rather than a countdown.
Beside how long each one lasted, another pattern is worth drawing out. Each of the three divided into two phases, separated by a fall of 20% below the price at which the regime began. That threshold is not arbitrary. At Bitcoin's realised daily volatility a 10% move sits inside a single month's normal fluctuation, while 20% marks a dislocation the market has to acknowledge.
Phase A, the range-bound phase, ran a different length each time, and in all three the break of that floor fell within days of a discrete structural failure:
Phase B carried the deepest damage in every case, and that sequence has shrunk with every cycle. A larger and more liquid market absorbs distress and surrenders less when it finally breaks.
Today's market sits — against the price at which it entered the regime, having traded no lower than — across — days, and that span is shorter than the shortest Phase A on record. In this phase the dominant cost of holding is time rather than price, and a market held in that condition has historically grown more exposed to an outside shock rather than less. Any future transition into Repair requires the newest money to reclaim the Short-Term Holder cost basis and hold above it for more than roughly a week. The longer that transition is deferred, the more the balance of probabilities tilts toward the second phase rather than away from it.
The prior section argued that the dominant cost of this phase is time rather than price. That is not a matter of interpretation, because the frustration it produces leaves a trace in the ledger.
What that trace can tell us is who is doing the selling and on what terms. Every dollar locked in falls into one of four buckets, long-term or short-term holder, at a profit or at a loss, and the relative share of those four is what identifies the cohort actually driving the activity.
Figure 5 sets them out, stacked to 100% on a 30-day average, so the reading is a composition rather than a total. As of today, of every $100 of gains and losses locked in by all sellers, — is loss booked by long-term holders, meaning investors who acquired their coins more than 155 days ago. That is the largest single band on the chart, and the two loss bands together account for — of all realised value. The market is settling well over half of its activity below cost.
More important than which cohort dominates, and by how much, is the direction that share has been moving. It stood near 14% in mid-January and had reached 31.4% by the day the regime began in June. It went on rising into late July, peaking at 35.8% on 22 July, and has eased back over the week since. That climb is the informative part, because price went nowhere across it. July has traded between $60,246 and $66,155, a span of under 10%. The escalation was not a reaction to falling prices. It was a reaction to prices that would not move.
It is worth being equally clear about what this is not. In the three prior capitulations the same share peaked far higher, at 56.9% in 2018, 60.2% in 2014 and 73.2% in 2022. At its own peak of 35.8% this cycle has not yet produced surrender on that scale.
So far the note has placed the market in the first half of a typical capitulation, where the cost being paid is time rather than price, and identified the sellers in the driver's seat as long-term holders booking losses, most of them buyers of the last cycle top. What is left is to size that pressure, because the volume of those losses is what says how hard it is actually running.
Figure 6 counts the coins long-term holders sold below their cost basis each day, smoothed on a 30-day average. It is denominated in coin rather than dollars deliberately. A dollar figure grows with the price and with the network, so 2015's capitulation and this one are not comparable in USD, whereas in BTC they measure the same unit of surrender.
The series sat at nothing through the October 2025 top, when almost no long-term holder was underwater, and turned up in November. It has not climbed smoothly since. It has stepped in waves, each higher than the one before:
The floors between those waves have lifted as well, from around 1.0k BTC a day in January to roughly 2.1k in May, which is what confirms the frustration is still growing with time, and the volume it forces out with it.
The 2018 and 2022 capitulations, measured on the same 30-day average, both peaked above 32k BTC a day. This cycle's highest reading to date is 17% of that, and the current pace nearer 11%.
When a capitulation ends is not forecastable, and this framework does not attempt it. Nor does it claim this cycle must reach the surrender of prior bears before it can turn. The purpose is narrower and more useful, which is to define a condition precise enough to recognise the cool-down when it arrives rather than to guess at its date. Any transition into Repair requires this pressure to dissipate, and only two routes lead there. An extended range can digest it slowly, absorbing top-buyer supply as patience gives out in instalments. Or a single wave of large loss realisation can clear it at once, which in every prior episode arrived with an external shock rather than from within. Neither route is currently favoured by the data, and the two map onto the same phases described earlier.
Run the framework end to end and the same picture assembles at every step. The market has been in Capitulation for — days, the weakest of the six orders. Against completed episodes of 283, 205 and 208 days, that is early.
Every prior capitulation ran in two acts. The first, Phase A, is a range. Price does not collapse so much as refuse to go anywhere, and the cost of holding is paid in months rather than in drawdown. Those stretches lasted 96, 71 and 141 days. The second, Phase B, begins when the floor gives way by 20%, and it carries the real damage. In all three cases that break came within days of something external breaking first, the Bitstamp breach, the hash war, the collapse of FTX. This episode is still in the first act, and has not broken.
That is where the frustration does its work. When price will not move, the people who paid the most are the ones whose patience runs out first, and the ledger names them. Long-term holders selling at a loss are now the largest single band of realised value. Their pace has done neither of the two things that would settle the question. It has not reached the surrender of 2018 or 2022, running near — a day against peaks above 32k BTC. Nor has it cooled. The floors between its waves have risen instead.
So the market sits in the part of the story where nothing has broken and nothing has resolved, which on a sample of three is the part that has ended with a shock rather than with a recovery. The alternative is available and it is specific. The newest money would have to reclaim the Short-Term Holder cost basis, still — percentage points away, and hold above it into Repair while long-term holder selling cools with it. Until that happens, more time is the base case, and the risk around it is not symmetric.