Supply and Scarcity · Published

Bitcoin Supply Shock Stalls: ETFs Returned 6,539 BTC

Over the last ten reported sessions, US spot Bitcoin ETFs absorbed none of Bitcoin's new supply. They returned 6,539 BTC to the market while miners created 4,500, an absorption ratio of -1.45x. That is the clearest sign in months that the bitcoin supply shock has stalled, and it sits underneath a twenty-session headline figure that still reads a comfortable 2.14x only because a single strong week in late September is still inside the window.

How much bitcoin did ETFs absorb against new supply over the last ten sessions?

The absorption ratio converts each session's net ETF flow into BTC at that session's own close, then divides by the Bitcoin mined over the same sessions. Above 1.0, the funds took more coin off the market than miners created. Below zero, redemptions handed coin back.

Across the ten sessions from 29 September to 12 October, the twelve US spot funds were net sellers of 6,539.3 BTC against 4,500 BTC of issuance, or -1.45x. The last five sessions are worse: -7,963.4 BTC against 2,250 mined, a ratio of -3.54x. For every single Bitcoin miners produced in the last week, the ETF channel released roughly three and a half.

Of the last five sessions, four were net negative:

The 7 October session alone gave back thirteen days of mined supply. On 12 October, the most recent session, none of the twelve reported products took in money.

Why the headline bitcoin supply shock reading still prints 2.14x

The twenty-session window currently runs from 15 September to 12 October, and over it the funds absorbed 19,216.7 BTC against 9,000 BTC mined, a ratio of 2.135x. That is the number the site scores on the institutional signal dashboard, and read on its own it badly misdescribes the present.

Split the window in half and the reason is obvious:

The entire positive reading is the first half. The second half is a drawdown. This is the structural weakness of any trailing window, and a reader who quotes 2.14x as a description of present ETF demand versus bitcoin mined is quoting late September.

Last week's base effect arrived, and demand overwhelmed it

Last Tuesday's supply absorption read made a specific, checkable forecast, and it is worth marking honestly. The reading then was 2.58x, and the five oldest sessions in that window, 8 to 14 September, were a net -3,949.5 BTC. Those sessions were about to roll off. The article said that if the next five sessions were completely flat, the ratio would mechanically climb to roughly 3.02x on the base effect alone.

The base effect arrived exactly as described. The ratio still fell, to 2.135x. The arithmetic reconciles precisely:

The roll-off was worth about +0.44 turns. The incoming week was worth about -0.89. New redemptions beat the favourable base effect by roughly two to one, which is the single most informative fact in this week's data.

2026's bitcoin supply absorption is net negative at -0.08x

Zoom out past the trailing windows and the year looks nothing like a supply squeeze. Across all 195 reported sessions in 2026, US spot Bitcoin ETFs absorbed a net -7,114.0 BTC against 87,750 BTC of issuance. The year-to-date absorption ratio is -0.08x.

Through more than nine months of 2026, the ETF channel has returned slightly more Bitcoin than it has taken in, while miners added 87,750 coins to the float. Whatever has supported the price this year, cumulative net ETF demand measured against new supply is not it. The bitcoin supply shock, as an ETF phenomenon, is a 2024 and 2025 story that has not repeated in 2026.

A 2.14x window inside a -0.08x year is noise around a flat trend, not the resumption of one.

Where -1.45x sits against the metric's own history

The series now runs 706 sessions, back to the first ETF session on 11 January 2024, which gives these readings real context.

The twenty-session range runs from -8.62x, set on 10 June 2026, to 11.82x in November 2024. So the present is not an extreme on any horizon: a middling headline covering a genuinely weak fortnight. Short-horizon bitcoin supply absorption has fallen into the bottom quintile of its own history while the scored figure has not yet noticed.

What bitcoin daily issuance of 450 BTC changes, and what it does not

Bitcoin's issuance is a step function. The series straddles the step: 72 of its sessions were mined at 900 BTC per day, and 634 at 450, with the first 450-per-day session on 22 April 2024. The bitcoin halving cut supply in half and mechanically doubled this ratio for identical demand.

It is worth seeing how much of the headline is the denominator. Measure this window's 19,216.7 BTC of absorption against the pre-halving rate of 900 BTC per day, and 20 sessions would have issued 18,000 BTC for a ratio of 1.07x rather than 2.14x. The same buying, against the old schedule, would read as barely keeping pace with miners.

But note what the halving does not do. It cannot change a sign. The last ten sessions are negative at 450 BTC per day and would be negative at 900, or at any issuance rate, because the numerator itself is below zero. Scarcity arithmetic flatters positive demand and offers nothing to redemptions. This is the part of the bitcoin halving supply argument that gets overstated: the halving doubles the ratio when funds are buying, and is irrelevant when they are selling.

At the latest close of $81,812, those 450 coins are about $36.8M of new supply per day. Over the last twenty sessions, valued at each day's own close, issuance totalled roughly $745M, about $37.2M per session. That is the practical answer to how much bitcoin do ETFs buy to stand still: a little under $37M of net inflow per trading day just to absorb what miners create, before anything is left over to tighten the float.

ETFs still hold 678,284 BTC, 70,291 below last October's peak

Cumulatively, since January 2024, US spot Bitcoin ETFs hold a net 678,284 BTC, worth about $55.5B at the latest close. Three ways to size it:

That stock does not vanish on a bad fortnight. But it is no longer growing, and the series records the turn. The cumulative figure peaked at 748,575 BTC on 9 October 2025. It now sits 70,291 BTC below that peak, a year on. Last week's read put the stock at 686,247 BTC; this week it is 678,284, the 7,963 BTC difference being exactly the last five sessions of redemptions.

So the flow is volatile and the stock is sticky, but sticky is not the same as rising. For the dollar-denominated side of the same story, including which issuers are driving it, see Monday's ETF flow read and the ETF flow dashboard.

Strategy added 334 BTC while the ETF channel was selling

Corporate treasuries compete for the same 450 coins a day. Strategy's most recent filing, dated 5 October, disclosed 334 BTC acquired, taking its holdings to 848,000 BTC; across its last four filings it acquired 2,949 BTC, with no disposals.

That 334 BTC is about three quarters of one day's issuance: a real second source of absorption, pointed the opposite way to the funds, but far too small to offset a 7,963 BTC ETF drawdown. Thursday's treasury watch covers the balance sheet behind it, and the filing data is eight days old, so a newer purchase may not be captured.

What next week's reading does even if nobody trades

The same base-effect check that failed to save the ratio last week now points firmly the other way, and it is worth flagging before anyone misreads next Tuesday's number.

The five sessions rolling out of the window next, 15 to 21 September, are a net +9,125.3 BTC — the strongest stretch in the current window by a wide margin. If the next five sessions were completely flat at zero net flow, the twenty-session ratio would fall from 2.135x to about 1.12x purely because that block of buying left the comparison. Any genuine redemptions on top of that would push the headline toward 1.0 or through it.

The headline is about to converge on what the ten-session reading has been saying for a fortnight. If it does fall next week, that will be a base effect first and fresh selling second, and the two should not be confused.

What this bitcoin scarcity measure cannot see

The absorption ratio is a demand proxy, and a narrow one. Its limits matter, because the number is easy to over-read in both directions.

It counts one channel. The numerator is US spot ETF flow only. OTC desks, exchange balances, sovereign buyers, miners' own treasuries, and every non-US fund are invisible to it. A -1.45x reading means the ETF wrapper was a net source of coin, not that the world was selling Bitcoin.

Its denominator understates mining. Issuance is counted at 450 BTC per reporting session, and reporting sessions are weekdays. Mining does not stop at the weekend. The current window spans 28 calendar days, in which roughly 12,600 BTC was actually mined against the 9,000 the metric counts — on a calendar basis this window is 1.53x, not 2.14x. Across the whole series the metric's denominator captures about 70% of calendar issuance. Every reading here, positive or negative, is scaled against trading days rather than real days.

A negative reading is redemptions, not overproduction. Below zero means shares were redeemed and coin was released. Miners did not out-produce anyone.

The price conversion cuts both ways. Flow is converted at each session's close, so an identical dollar inflow buys fewer coins as price rises. Closes in the current window ranged from $75,608 to $86,662, which is enough spread to move the BTC-denominated figure on its own.

It is not a price forecast. There is no mechanical link from this ratio to price, and this fortnight illustrates it: the funds were net sellers of coin for ten sessions without that resolving into any clean price outcome. Treat it as one input on the supply side of a market with many participants, which is how the composite on the institutional signal dashboard weights it, at 20% of a reading currently described as neutral or mixed.

The one thing this week's data does say clearly is this: the comfortable headline is carried by sessions that are about to expire, and underneath it, ETF demand against bitcoin mined has been negative for two weeks.