Bitcoin Supply Shock Rose to 3.8x as ETF Dollars Faded
Over the last 20 reporting sessions, from August 31 through September 28, US spot bitcoin ETFs absorbed 34,005 BTC while miners created 9,000. That is a bitcoin supply shock reading of 3.78x: for every coin issued, the funds took 3.78 off the market. The striking part is the direction. Last Tuesday this same rolling window read 3.22x, and every session since has delivered a smaller dollar figure than the one before it — yet the supply ratio went up, not down.
That apparent contradiction is the whole story of the week, and it is a good illustration of why the absorption ratio and the dollar flow headline are not the same measurement.
How much bitcoin did ETFs absorb against new supply this week?
Across the five sessions from September 22 to September 28, the funds took in $1,417.8M, which converted at each day's close works out to 16,630.7 BTC. Miners produced 2,250 BTC over the same five sessions, at the fixed post-halving rate of 450 BTC per day.
That is a ratio of 7.39x for the week on its own — more than seven coins absorbed for every one created. The week's demand, in other words, was running at better than seven times the rate at which new bitcoin scarcity is relieved by mining.
Why the bitcoin supply shock rose while the dollars shrank
A rolling 20-session window changes for two reasons: what enters it, and what leaves it. Monday's flow article covered the first half — inflows faded in five straight sessions, from $999.0M down to $134.5M. Read alone, that looks like a metric on its way down.
But the sessions dropping out of the back of the window were worse than the ones arriving at the front. The late-August and early-September stretch that rolled off included September 1 at -3,054.8 BTC, and the fading sessions that replaced it were all still positive. A shrinking inflow is still an inflow. Replacing a redemption day with a modest creation day lifts the average even when the newest number is the smallest in weeks.
This is the practical difference between the two readings. The dollar series answers "is demand accelerating or decelerating?" The absorption series answers "is demand still outrunning issuance?" Last week those two questions had opposite answers, and both answers were correct.
Monday's session bought less than one day of bitcoin daily issuance
The one genuine warning in the data is the newest session. On September 28 the funds took in $31.0M, which at that day's close of $83,457 is 371.45 BTC — against 450 BTC mined. That is a ratio of 0.825x, the first session since September 16 in which ETF demand failed to cover a single day's issuance, ending a run of seven consecutive sessions above 1.0x.
One session is not a trend, and the seven that preceded it ranged from 3.56x to 25.62x. But it is the first hard evidence that the dollar fade has gone far enough to matter in coin terms rather than only in percentage terms.
Where 3.78x sits against its own history
The rolling series now runs to 677 readings. The current 3.78x sits at the 68th percentile of that full history — elevated, but well short of the 11.82x peak set in November 2024.
Measured against 2026 alone, it looks much stronger. This year the same rolling window has a median of just 0.363x, and 87 of its 185 readings have been negative, meaning the funds were net returning coins to the market rather than removing them. Only 79 of 185 readings cleared 1.0x at all. Against that backdrop, 3.78x ranks at the 87th percentile of 2026.
Some perspective on what that ranking does not mean: this is not a 2026 high. The window read 4.06x as recently as September 15, and peaked this year at 5.25x on May 6. The correct reading is that absorption is firmly in the upper band of a year that has mostly hovered around zero, not that it has broken out of it.
What the halving did to this arithmetic
Bitcoin daily issuance is a step function, and the April 2024 halving cut it from 900 BTC per day to 450. Nothing about demand changed on that date — the denominator simply halved.
The effect on this metric is exactly proportional, and worth stating plainly because it is easy to over-read a rising ratio as rising demand. The same 34,005 BTC absorbed over the same 20 sessions, measured against the pre-halving issuance rate of 900 BTC per day, would produce a ratio of 1.89x instead of 3.78x. Identical buying, half the reading.
That is the mechanical core of the bitcoin halving supply argument: a fixed, scheduled reduction in new supply means any given level of demand clears a progressively larger share of it. The next halving will do the same thing again. It is also why comparing today's ratio against readings from before April 2024 requires care — the 2024 highs in this series were achieved against a denominator twice the current size.
ETFs now hold 1.98x everything mined since January 2024
Since the series begins on January 11, 2024, cumulative net absorption stands at 684,823 BTC. Over those same 696 sessions, miners produced 345,600 BTC. The funds have therefore taken up 1.98 coins for every coin created across the entire life of the US spot ETF complex.
Two framings of that 684,823 BTC:
- It is 3.26% of bitcoin's 21 million hard cap, held by a vehicle class that did not exist before January 2024.
- At the current 450 BTC per day, it represents 1,522 days of issuance — roughly four years and two months of future mining, already absorbed.
The second figure is the one that actually describes the scarcity mechanic. It is not a forecast; it is a statement about coins that have already changed hands.
Strategy added 1,665 BTC on top of that
ETF flow is not the only institutional bid competing for the same 450 coins a day. Strategy's September 28 filing disclosed 1,665 BTC acquired for $142.7M at an average of $85,681, taking total holdings to 847,666 BTC.
Adding that to the week's ETF absorption gives 18,295.7 BTC against the same 2,250 mined, or 8.13x. The period alignment is approximate — the filing covers Strategy's own reporting week rather than precisely the five ETF sessions — so treat it as a scale comparison rather than a combined metric. Thursday's treasury article is where that balance sheet gets proper attention.
What this bitcoin scarcity measure does not capture
The absorption ratio is a clean number, which makes it easy to ask more of it than it can bear. Four limits worth holding onto:
- It counts US spot ETF flow only. OTC desks, exchange balances, sovereign holders, and every corporate treasury other than the ones tracked separately are invisible to it. Real demand for the daily 450 is higher than this ratio shows, by an amount nobody publishes.
- A negative reading means redemptions, not overproduction. When the ratio went to -8.62x on September 16, miners did not out-produce anyone. Investors sold ETF shares, and the funds released coins. The denominator never goes negative.
- It is a demand proxy, not a price model. A high absorption reading tells you the funds outbid new supply. It does not tell you what price does next, and this series contains plenty of stretches where a strong ratio preceded a flat or falling tape. The rolling window read above 3x on 45 separate days this year, in both directions of price.
- Converting dollars to coins at each day's close introduces a real distortion. 2026 is the sharpest illustration available. Year to date the funds have taken in a net $1,052.3M in dollar terms — but converted daily, that is net -574.6 BTC. The reason is price asymmetry: 2026's inflow days converted at an average of $75,823 while its outflow days converted at $72,273, so redemptions released more coins per dollar than creations absorbed. A positive dollar year and a flat-to-negative coin year are both true descriptions of the same 185 sessions.
That last point is the one to carry into next week, because it cuts both ways. Absorption measured in coins is the more honest scarcity measure, and it is also the more volatile one.
What would have to happen to hold 3.78x
The mechanics that lifted the ratio this week work in reverse from here. The five sessions due to roll out of the 20-session window carry 12,198.6 BTC between them, including the 8,995.1 BTC absorbed on September 3. The remaining 15 sessions carry 21,806.5 BTC.
If the next five sessions absorbed nothing at all, the window would still read 2.42x. To hold 3.78x, the funds need to replace the full 12,198.6 BTC — an average of 2,439.7 BTC per session, or roughly $203.6M per day at Monday's close.
Last week ran at 3,326.1 BTC per session, comfortably above that bar. Monday's 371.45 BTC was not. The gap between those two figures is the entire question for the week ahead, and the rolling ratio is drawn daily on the institutional signal dashboard, with the underlying dollar series on the ETF flow dashboard.
Every figure above is computed from bryptoflows' own absorption series, which converts each reporting session's net flow at that session's close and divides by the 450 BTC mined over the same period. Last week's supply absorption read covers the single-session record that anchors the current window.