Supply and Scarcity · Published

Bitcoin Supply Shock Reverses: ETFs Hand Back 3,750 BTC

Over the twenty reporting sessions through October 26, US spot Bitcoin ETFs returned 3,750.4 BTC to the market while miners created 9,000 BTC — an absorption ratio of -0.42x. The bitcoin supply shock this column has been tracking has not merely stalled; for the first time since late July it has inverted, with the largest institutional buyer channel acting as a net supplier of roughly 8.3 days of global issuance.

That also settles a specific bet this column placed last week. Seven days ago the twenty-session reading was +1.61x, and the arithmetic said the funds needed 11,145 BTC over five sessions to hold the headline at 1.0x. They delivered -1,605.1 BTC. The hurdle was missed by more than the hurdle itself.

How much bitcoin did ETFs absorb against new supply this week?

Every figure here is computed from the absorption series behind the institutional signal dashboard, which converts each session's net ETF flow to BTC at that session's close and divides by the Bitcoin mined over the same sessions. The series runs 716 reporting sessions, from January 11, 2024 through October 26, 2026, and the API's own supply-absorption block agrees to four decimals: -3,750.4 divided by 9,000 is -0.4167.

The five windows that matter do not agree with each other:

Read down that list and the sign alternates four times. That is the structure of bitcoin supply absorption right now — buying that arrives in violent single-day bursts, inside a quarter that was in aggregate strongly accumulative. Anyone quoting one of those five numbers as "the" absorption figure is choosing their conclusion before their window.

Two adjacent sessions moved 31 days of issuance in opposite directions

The week's internal shape explains why the windows disagree. Session by session, in coins rather than dollars:

October 21 and October 22 are consecutive sessions. Between them they moved 31 days of issuance in opposite directions and netted -1,562.8 BTC — close to nothing, after a round trip that would be an eventful month for the mining network. The week's -1,605.1 BTC is almost entirely those two days cancelling, which is why the five-session reading sits at only the 33rd percentile rather than somewhere extreme.

Monday's rebound carries a note the dollar figure hides. The entire +$110.3 million, and so all 1,376.9 coins, came through a single fund: BlackRock's IBIT reported the full amount and every other product reported exactly zero. bryptoflows records an unreported product as null rather than zero, so those are genuine zeros — three days of global issuance left the market through one wrapper.

Where this bitcoin supply shock reading sits in its own history

The ratio is unbounded, so the level means little without its distribution. Across all 697 rolling twenty-session readings, -0.42x sits at the 27th percentile, against a median of +2.07x. The range runs from -8.62x on June 10 this year to +11.82x on November 21, 2024, during the post-election creation wave. This is a weak reading, in the bottom third but nowhere near the worst on record.

Against 2026 alone it is less unusual still. This year's 205 rolling readings have a median of just +0.83x, and 43.9% are negative, putting today's -0.42x at the 40th percentile. A negative twenty-session reading is close to a coin flip in 2026 — the more durable finding, because the bitcoin scarcity story this channel supplied through 2024 and much of 2025 is not what this year's data describes.

The speed is the uncommon part. The reading fell 2.03 points in five sessions, from +1.61x to -0.42x; only 14.6% of the 692 five-session changes on record are steeper declines, the record being -5.97 on November 4, 2025. And almost none of the drop came from this week's trading — the five strong late-September sessions rolling out of the back of the window did the bulk of it.

Why the next reading is -0.58x if flows do nothing

The same arithmetic runs forward, and this time it cuts the other way. Over the next five sessions the window loses September 29 through October 5, which contributed +1,424.1 BTC between them. What remains is the fifteen sessions from October 6 to October 26, summing to -5,174.6 BTC. If the next five sessions produce exactly zero net flow, the reading published next Tuesday is -5,174.6 divided by 9,000, or -0.58x.

But the hurdles have collapsed against last week's, because the window is no longer measured against September's best stretch:

Last week the bar for 1.0x was $183 million a session and the channel delivered negative flow. This week the bar for getting back to zero is $83 million a session, which the sixty-session average clears comfortably. The honest reading: the headline will probably look worse before it looks better, and that will be a statement about September's roll-off, not about the next five days.

One event sits inside that stretch: the Federal Reserve meets October 27 and 28, with the decision due Wednesday afternoon. Flow data lags by a day or more, so any reaction lands in Thursday and Friday prints.

Bitcoin daily issuance is 450 BTC, and nothing changes it

The denominator is why this metric is worth computing. Bitcoin daily issuance has been fixed at 450 BTC since the April 2024 halving cut it from 900, and stays there until the next halving in 2028 takes it to 225. It does not respond to price, ETF demand, hashrate or miner profitability: more competition between miners produces the same coins at higher cost.

That fixed step is what makes the comparison meaningful, because every other bitcoin scarcity measure rests on behavioural assumptions about holders, exchange balances or lost coins, and the issuance line needs none. The bitcoin halving supply schedule does mean cross-era readings need care: the first 69 sessions in this series ran at 900 BTC a day, so a 5x in early 2024 and a 5x today describe the same ratio against twice the absolute supply.

2026 is up $767 million in dollars and down 4,325 BTC in coins

This finding only appears when you measure in coins, and this week widened it. Across 205 reporting sessions in 2026, US spot Bitcoin ETFs have taken in a net +$766.6 million while absorbing -4,325.1 BTC — a year positive in dollars and negative in bitcoin at once, for a ratio of -0.05x against 92,250 BTC mined.

The cause is the conversion price. Dollars arrived on inflow sessions at a flow-weighted average of $76,325 a coin, buying 330,496.5 BTC with $25,225.3 million. Dollars left on outflow sessions at $73,050, taking 334,821.6 BTC out with $24,458.7 million. Money came in when Bitcoin was dearer by about $3,275 a coin, or 4.5%, and went out when it was cheaper, so near-identical dollar totals moved materially different quantities of Bitcoin. A fund complex can report a positive year in its reporting currency and still have supplied coins on net. October alone runs -$203.2 million and -2,763.6 BTC across eighteen sessions, a ratio of -0.34x.

ETFs hold 681,073 BTC, and absorption alone is dragging the composite

Cumulatively, the ETF channel has absorbed 681,072.9 BTC since January 2024 — 3.24% of the 21 million coins that will ever exist. Against the 354,600 BTC mined across the same reporting sessions that is 192%, but the series counts weekdays only while mining runs every day. On a calendar basis roughly 504,000 BTC have been issued since January 11, 2024, so the channel has taken about 135% of new supply — the better answer to ETF demand vs bitcoin mined as most readers mean it.

The cumulative total peaked at 748,575 BTC on October 9, 2025 and now sits 67,503 BTC below that mark. Whatever the ETF channel is, it is not a ratchet.

What this reading does to bryptoflows' own composite is unusually clean. The institutional signal scores -9.8 on data through October 26, labelled neutral or mixed, with all eight components live — an improvement on the -22.1 that Friday's signal column reported. Within it, supply absorption normalises to -76.7 and contributes -15.3 of the score, while the other seven components together contribute +5.5. Absorption is not one input among eight here; it is the entire reason the composite is negative.

Strategy bought nothing, so both absorption channels went quiet together

Corporate treasury demand is the other institutional bid on new supply, and it filed on Monday. Strategy's 8-K of October 26 reports direction "none" and a btc_change of zero: no purchase, holdings unchanged at 848,517 BTC, cost basis unchanged at $64.02 billion, or $75,446 a coin. It is the first zero-purchase filing since September 14, ending a run of six acquiring weeks, though such filings are not rare — twelve of the seventy-one on record carry direction "none".

Strategy's last four filings total 851 BTC against roughly 12,600 BTC of calendar issuance over those four weeks, about 6.8%. So in the week the ETF channel's twenty-session reading went negative, the other institutional absorption channel contributed nothing. Thursday's treasury column covers what that means for the company; for the supply arithmetic it means no offset.

What this bitcoin scarcity measure cannot see

The absorption ratio is precise about a narrow thing, and the limits need restating every week.

For the dollar-level view of the same tape, Monday covered ETF flows and October's month-end maths, and last Tuesday's absorption read set the hurdle this week failed to clear. Live figures are on the ETF flow dashboard and the institutional signal dashboard, where the absorption chart is drawn.

All figures are computed from bryptoflows' own read-only API — ETF flows from Farside Investors, prices from Hyperliquid, Strategy detail from SEC filings — current through the October 26 close and 8-K. Published October 27, 2026.