The decision

What changed this quarter. Price rose 43%. ETF flows turned positive on the year. Miner economics moved off the left arm of the U-curve to the trough, while miner participation turned back down. The correction did not complete, and the alternative is the case that tracked price. $87,396 became the level that most moves the balance between the two.

Everything that follows is the evidence for those lines, the limits of that evidence, and the observations that move the weighting between the two.

On precision. Market observations are given to the nearest dollar where the source supports it. Structural levels are exact framework outputs and are stated without qualification. External estimates — mining costs, holder shares, corporate holdings — are approximate and are marked as such.

Where the quarter ended

Bitcoin closed September at approximately $83,712, against a second-quarter close of $58,524 — a gain of roughly 43%, the strongest quarterly result since the first quarter of 2024. The asset remains approximately 34% below its all-time high of $126,296 and approximately 441% above the $15,460 cycle low used as the prior-cycle reference throughout this series.

Two low references are now in use and they should not be mixed. $15,460 remains the prior cycle low and the anchor for long-horizon gain figures. $57,800, set on 1 July 2026, is the 2026 low. The alternative case set out below treats the July low as this cycle’s low; the base case does not. Where a figure in this edition refers to a gain from the low, the reference is stated.

The Q4 map

Every level below is an area of interest. Two of them are decisions, and they are the only claims in this table.

On $87,396. It is the September high, and it matters because of what trades above it rather than because it is a round number. A non-overlapping five-wave advance through it would remove the simplest corrective readings of the move from July. It would not remove all of them. Several corrective structures require a motive five-wave advance in this position — the C wave of a flat is the clearest case — so a five-wave advance is evidence against the correction thesis rather than proof against it.

What it does is shift the weight. Above $87,396 on that structure, this house would hold a mid-term bullish position with the longer-term correction still intact, and the burden of the base case would be visibly heavier. The level is therefore reported as a probability shift, not as an invalidation, and this edition does not claim it as one.

The distinction matters. Whether the market trades above $87,396 is an observable fact. What a qualifying five-wave structure through it does to the correction thesis is a judgement of the house framework, not an independently measured one.

What the record shows

Alderlux publishes its calls and reports back on them. Three things from the September record belong in this edition before anything forward-looking.

The Q2 correction thesis is delayed, not invalidated. It held the correction unfinished with a low in the first half of 2027. A quarter of upward price action does not disprove it: the timeline it describes has not yet arrived, and the desk’s largest-degree structural primary remains the bearish count. It is the Q2 expansion thesis — a low in the second half of 2026 followed by expansion before year end — that has tracked price over the quarter. It is carried below as the alternative, with its evidence in full.

The distinction matters and is stated precisely. A view that is delayed is still a view. A non-overlapping five-wave advance through $87,396 materially lowers this thesis without retiring it, because corrective structures exist that require exactly that advance in this position. It is reported as a probability shift and nothing stronger.

The mid-$50,000s call died at $82,850 on 21 September. The call as published is retired, and it is not quietly reframed.

What is retired is that formulation and its condition. It is not the expectation of lower prices, which this house still holds: the base case carries an objective near $57,000 in the first half of 2027. The two are separable because they were made at different degrees and on different conditions — the September call carried a stated level that was reached, and the base case objective does not sit behind it.

Nothing in this edition should be read as a view that price goes up from here. Current conditions do not favour the lower outcome; the structure continues to lean towards exactly that outcome. Both of those are true at once, and the weighting below is what reconciles them.

In the first-quarter edition, with Bitcoin near $120,000, this house identified the $54,000 region as a downside target. The 2026 low printed at $57,800. That is within approximately 7% of the eventual low, called from roughly double the price. Both belong on the same page: the call that landed, and the call still waiting on its condition.

The two cases

Base case: the Q2 correction thesis remains alive. The correction is unfinished, with the objective near $57,000 and a first-half 2027 timeframe.

The reasoning is structural rather than price-based, which is why a 43% quarter does not disturb it. The desk’s largest-degree structural primary remains the bearish count, and the timeline the thesis describes has not yet arrived. A view that has not yet reached the period in which it expects to resolve has not been tested by the market.

A non-overlapping five-wave advance through $87,396 materially lowers this base case. Below that level, the simplest corrective readings of the advance from $57,800 remain available under the house wave rules. Above it, on that structure, those readings narrow to the corrective patterns that require a motive five-wave advance in this position — fewer paths, and a heavier burden on the thesis, but not its end. In that event this house would hold a mid-term bullish position with the longer-term correction intact.

Alternative case: the cycle low formed at $57,800 on 1 July 2026, and the advance from it is the first leg of what the house framework classifies as an expansion.

This is the case that has tracked price through the quarter, and it is given its evidence in full rather than dismissed because it is the alternative.

Taken together these support the expansion case without proving it.

This alternative retires on a daily close below $57,800. The case rests on the July low being the cycle low. If price trades through it, the premise is gone, and no reinterpretation of the advance above it survives that.

On $72,150. The daily VWAP base of the advance from the July low is not an invalidation of either case. It is the phase boundary: below it, this framework no longer classifies the advance from July as a recovery. That matters for how the move is described and for what the next edition would be reporting, but it ends neither thesis. It is included because a level that changes the classification without changing the position is exactly the kind of thing that gets misread as a trade trigger.

A note on the two properties above, which are deliberately separated. Conviction is how strongly this house holds the view: medium. Falsifiability is whether the view can be ended by something observable: high, because both cases carry a stated structural condition. The previous edition’s single “Medium-High” conflated the two and was attached to nothing.

On this edition the Insider’s base case and the desk’s largest-degree structural primary agree — both are the bearish count. They have differed before and may differ again; where they do, it will be said here rather than left for a reader to discover.

The evidence hierarchy

This series separates what is measured from what is judged. From this edition the separation is explicit.

The rule: a Tier 3 conclusion cannot override a Tier 1 deterioration without a new piece of evidence. If flows, open interest and mining economics turn while the structural reading still says expansion, the structural reading is the thing that gives way. The base case above is a Tier 3 conclusion resting on Tier 1 and Tier 2 evidence, and it is held on that basis.

The hike produced volatility, not direction

On 16 September the Federal Reserve raised its policy rate by 25 basis points to a target range of 3.75%–4.00%, the first increase since 2023. The dot plot points to a median of 4.1% at year end. Bitcoin then closed its strongest quarter in more than two years.

The hike mattered to rates expectations, yields, the dollar and financial conditions. What the house record addresses is narrower and is the only claim made here: whether a Federal Reserve decision is followed by a persistent directional move in Bitcoin on the measured timeframe. It is not.

This house maintains a standing record on rate decisions. Across sixteen prior events, plus this one, Bitcoin’s hourly range following a Federal Reserve decision has typically been two to three times a normal hour’s range for approximately four hours, with no consistent directional tendency across the sample. The 16 September bar did exactly that: 1,260 points of range, net flat on the session.

The subsequent advance occurred alongside two other identifiable changes: the failure of the Clarity Act vote the previous night, and the unwinding of leverage that had accumulated against the shelf at $75,000. The 12% move over five sessions is not attributed to any of the three here. They are named for one reason: to show that the decision alone is not a sufficient explanation of what followed. Attribution between them would require a test this edition has not run.

Method. Event-window analysis of seventeen Federal Reserve decisions, comparing Bitcoin’s realised hourly range and directional return in the four hours following each decision against the preceding baseline distribution. “No consistent directional tendency” means none was observed across the sample; it does not mean individual decisions produced no directional moves. The sample has not been tested for statistical significance and has not been re-run excluding extreme observations. It is a finding about volatility, held at that strength and no higher.

Flows and holders

ETF flows turned, and the turn is measurable. US spot Bitcoin ETFs took in a net $6.36 billion across the third quarter, against a net outflow of $4.89 billion in the second. The year-to-date figure crossed back into positive territory and closed the quarter at +$970 million, having bottomed at −$5.69 billion on 13 July.

The shape matters more than the total. Q2 had 23 positive sessions out of 62; Q3 had 42 out of 64. This was persistent demand across the quarter rather than a handful of large prints, and the largest single session — $999 million on 21 September — accounts for under a sixth of the quarterly total.

Two qualifications. The quarter’s final session, 30 September, was an outflow of $149 million, so the streak that ran from 17 September ended inside the quarter. And total net assets of approximately $108.4 billion are a function of price as well as flow; they rose in Q3 for both reasons and the two should not be read as one.

Corporate holdings. Public companies hold approximately 1,275,684 BTC — roughly 6.08% of total supply — across 197 companies, per BitcoinTreasuries.net as at 1 October 2026.

From this edition, this series reports corporate adoption by Bitcoin held rather than by company count. The count moves when companies delist, merge, are reclassified or fall to zero holdings while remaining listed; holdings measure the thing the metric is for.

Quarterly change: not stated. The prior series published the company count, so there is no comparable holdings figure for Q3. This edition establishes the baseline and quarterly change will be reported from Q1 2027.

Long-term holders. The share of supply held by long-term holders reached approximately 80%, an all-time high, having risen from approximately 65% over the preceding year.

Leverage and the marginal dollar

Two measured questions this edition can now answer: whether leverage has reset, and what a dollar of ETF flow is worth.

Leverage has unwound from the peak, not reset. Aggregate open interest across BTC perpetual markets peaked at $29.5 billion on 21 September — approximately 340,000 BTC — and stood at $25.9 billion, approximately 310,000 BTC, on 1 October. That is 12% lower in dollars and 9% lower in Bitcoin terms.

Against longer baselines the picture reverses:

Open interest also rose across the quarter as a whole, from $20.3 billion to $25.7 billion. “Reset” is therefore true only against the September high and false against every longer baseline. The term is retired from this series; the measure reported from here is the level against a stated baseline, and the baseline is named each time.

The same-day ETF and price relationship is stronger than it was. Regressing Bitcoin’s daily return on the same day’s net ETF flow gives approximately +3.3% per $1 billion across the full record since January 2024. By year: +3.7% in 2024, +2.1% in 2025, +5.3% in 2026, and +6.0% across the third quarter. The estimated same-day relationship is strongest in Q3 within this record.

The association is unlikely to be explained by a random permutation of the flow series. Shuffling the flow series against returns 2,000 times reproduced a slope this large in none of the draws.

Method. The permuted statistic is the regression slope; the comparison is two-sided, on absolute value. The specification was fixed before the test was run, with no controls. Observations are treated as exchangeable, which ignores serial dependence in both series — a real limitation, and the reason this is reported as a permutation result rather than a significance test. Because both variables are measured over the same trading day, the regression cannot establish whether flows led price, price led flows, or both responded to something else.

And it has no predictive content. The same regression run forward gives +0.4% per $1 billion against the next day’s return, with an R² of 0.003 — indistinguishable from nothing. Run backward, against the previous day’s return, it gives +3.0%, nearly as strong as the same-day figure.

The honest reading is that ETF flow and price move together within the day, and flow follows price at least as much as it leads. In this specification, a flow number tells you what has already happened, at a higher resolution than it used to. It does not tell you what happens tomorrow. For a member, that means the headline weekly flow figure is a description of the quarter, not an input to a decision about the next one.

The miners

Two miner readings now point in different directions, and this edition states that rather than resolving it.

Economics have improved. Hashprice stands at $42.20/PH/day. Against a revised Cost of Hash Threshold of $38/PH/day, the ALX·HSI reads +11% — the Moderate Profit zone, the trough of the U-curve, where selling pressure from the median operator is at its minimum. Q2 read approximately −11%, At Breakeven, on the left arm of the curve where forced selling is elevated.

The threshold was revised this quarter, from $35 to $38/PH/day, following CoinShares’ Q2 2026 report. That report puts listed miners’ weighted-average ex-tax cash cost at approximately $75,500/BTC, which at current network conditions implies a breakeven near $37.95/PH/day; the same report describes hashprice around $38 as the level returning most operators to cash breakeven. Readings from this edition forward are measured at $38. The Q2 reading of −11% was measured at $35 and the two are not directly comparable.

Miner participation has not followed. The Hash Ribbon — the 30-day hashrate moving average against the 60-day — has moved back into capitulation. It read recovering earlier in the review period and has since crossed back. Hashrate through September was volatile enough to make that plausible, ranging from approximately 821 to 1,070 EH/s inside a fortnight.

These two readings are not contradictory, but they are not the same thing. The HSI measures whether the median operator is above the cost of production. The Hash Ribbon measures whether hashrate is being switched on or off. An operator can be above cash breakeven on the day and still be shutting machines down — because capital decisions respond to expected returns and fleet economics rather than to one day’s margin, and because the AI and HPC transition continues to pull capacity off Bitcoin mining regardless of hashprice.

The punchline. Economics improved; participation deteriorated; neither independently predicts price. Improved economics are evidence against forced selling by the median operator, not evidence of future appreciation — the trough is where selling pressure is lowest, not where accumulation begins. The Hash Ribbon’s association with cycle lows is a pattern across a small number of events rather than a tested edge. The second of the two readings is the newer information.

What would change the view

This house ranks scenarios rather than holding one and waiting for it to be disproved. A single level that ends a view is the exception, not the method: most of the time what changes is the weight carried by each case, and tracking that movement is more useful than waiting for a binary. The table below is therefore the operative part of this edition. Each line states which case it favours and why, and the next edition reports which of them occurred and how the weighting moved.

The view from the edge

A house that publishes a directional view will be wrong in public. The question is what it does next.

The Q2 correction thesis has not been invalidated. It has been delayed, and it has lost ground: the alternative is now the better-supported case, which it was not a quarter ago. A non-overlapping five-wave advance through $87,396 would lower it further. That level is reported for what it is and not inflated into an invalidation it does not deliver, because a condition claimed too strongly is worth less than a weaker one stated accurately.

What was settled is the mid-$50,000s call in the form it was published. Its condition was reached, and that formulation is retired. That is the system working — not because being wrong is comfortable, but because a view with a price attached can be settled, and a view without one cannot.

Settling that call is not a move to a bullish position, and this house is not claiming one. The expectation of lower prices stands. What has changed is the weight it carries now: current conditions do not favour it, while the structure continues to lean towards it.

What is published here is a weighting, not a prediction waiting to be graded. The alternative leads on current evidence; the correction thesis is carried behind it; the observations that would move either are set out in full, and a daily close below $57,800 ends the alternative outright. The next edition opens by reporting which of those observations occurred and how the weighting moved — including, and especially, when it moved against the house.

Levels are areas of interest, not trade instructions.

Basis and limits

Price, flow, open interest, rate and corporate holdings figures in this edition are drawn from public sources as at 30 September and 1 October 2026, and are stated to the precision the sources support. Open interest is summed across every BTC perpetual market tracked by Coinalyze, denominated in USD, daily close of bar; it excludes dated futures and options. The flow-response regression uses 698 trading sessions from 11 January 2024, with ETF flows from Farside and price from Coinbase spot. It is an association, not a causal estimate, and the forward-looking test reported alongside it is the reason the association is not presented as predictive. ETF figures are as reported by SoSoValue and Decrypt trackers; corporate holdings from BitcoinTreasuries.net; rate decisions and projections from the Federal Reserve.

The rate-decision record referenced above covers sixteen prior events plus the September 2026 decision, measured on intraday range and direction. It is a finding about volatility, not about price direction, and it is not a forecast.

The ALX·HSI reading, hashprice and Hash Ribbon state are taken from the live indicator, which sources network hashrate from Glassnode and difficulty from Quandl. The Cost of Hash Threshold was revised from $35 to $38/PH/day for Q4 following the CoinShares Q2 2026 mining report; readings before and after that revision are not directly comparable, and the revision alone reduced the reading by approximately nine percentage points without any change in market conditions. The Hash Ribbon state is as at the date of the indicator reading and crossed during the review period; a state that has recently crossed is more likely to cross again than one that has held.

Structural readings, phase classification and the levels above reflect this house’s framework and judgement. They are not measurements, and they are distinguished from the measured figures throughout.

Alderlux provides education, research and succession planning services in relation to Bitcoin and digital assets. We do not provide regulated financial advice, investment recommendations, legal or tax advice. Alderlux is not authorised by the Financial Conduct Authority and does not provide cryptoasset custody, dealing or arranging services. Bitcoin, cryptocurrencies and digital assets are speculative and volatile assets. Prices may rise or fall significantly within short periods of time. Investors may experience substantial losses, including the loss of their entire investment. Past performance is not indicative of future results.