Bitcoin Supply Shock: 11,528 BTC Taken in One Session
How many bitcoin did ETFs take off the market this week?
More than in any single session for almost two years. On Monday, September 21, US spot bitcoin ETFs absorbed a net 11,528 BTC while the network minted 450 — an absorption ratio of 25.6x, the sharpest one-day bitcoin supply shock since November 11, 2024, measured both in coins and as a multiple of issuance. Across the five reporting sessions from September 15 through September 21 the funds took 9,125 BTC against 2,250 mined, a ratio of 4.06x. That is the headline, and the rest of this piece is about why the trailing measure that the site actually charts went down in the same week.
The arithmetic is deliberately plain. Each session's net ETF flow is converted to bitcoin at that day's close, then divided by the 450 BTC of bitcoin daily issuance the network pays out post-halving. Above 1.0 the funds removed more coins than miners created; below zero, redemptions handed coins back. bryptoflows builds the series from Farside Investors flow data and Hyperliquid closes, and the rolling chart lives on the institutional signal dashboard, with the product-level dollar flows on the ETF flow dashboard. Everything below is computed from that series, current through the September 21 session.
Monday's session ranks sixth out of 691
Since the funds began trading in January 2024 the series holds 691 reporting sessions. Monday's 25.6x ranks sixth. The five above it are all from 2024: November 7 at 40.2x, June 4 and November 11 at 27.9x, October 30 at 27.4x and October 29 at 26.6x. Nothing in the 22 months since November 11, 2024 has absorbed either more coins or a higher multiple of issuance than Monday did.
Two things made it that large, and only one of them is demand. The $999.0M net inflow was the dollar figure, and Monday's flow piece covers how that broke down by issuer. The second factor is the halving. At the pre-April-2024 rate of 900 BTC per day, the identical $999.0M inflow at the identical $86,662 close would have printed 12.8x, not 25.6x. Half the headline number is the supply schedule, not the bid — which is precisely what the metric is built to show.
It is worth being clear about what a high single-day reading is not. It is not a claim that 11,528 coins left the market permanently, and it is not a price forecast. It is one session's net creations expressed in the only unit that makes bitcoin scarcity legible: coins, against the coins that came into existence alongside them.
The bitcoin supply shock reading still fell, from 5.24x to 3.22x
The twenty-session window is the one the chart draws and the one the composite score consumes. It now reads 3.22x — 29,017 BTC absorbed against 9,000 mined across August 24 to September 21. Last Tuesday it read 5.24x. A record-adjacent Monday did not stop it falling by two full ratio points, and the reason is roll-off.
Last week's piece set out exactly this and put a number on it. It observed that 27,228 coins — 57.8 percent of that window — came from five consecutive sessions in mid-August, that those five were about to age out, and that holding net flow at zero from there the ratio would decay to 2.21x after five sessions. Five sessions have now passed, so the forecast can be graded rather than repeated.
The mechanical part was exactly right. Those five August sessions rolled off and the window's carried-forward total fell to 19,892 BTC, which is 2.21x on the nose. The five new sessions then added their actual 9,125 BTC, lifting it to 29,017 BTC, or 3.22x. The gap between the 2.21x counterfactual and the 3.22x print is, to the coin, what the ETF complex genuinely bought in the last five sessions. Last week's forecast understated the outcome by a full ratio point, and every bit of that miss is real demand rather than arithmetic.
This is the discipline the metric demands. A trailing window moves for two reasons — what arrives at the front and what leaves the back — and quoting the level without separating them produces confident nonsense in both directions. Last week the back of the window was doing the work and the headline overstated demand. This week the front did more than expected and the headline still fell.
Where 3.22x sits against its own history
Across the 672 rolling twenty-session observations in the series, 3.22x sits at roughly the 61st percentile. The median reading is 2.01x and the mean is 1.77x. The extremes are 11.82x, set on November 21, 2024, and -8.62x, set on June 10 this year.
That places the current reading in unremarkable territory, which is the honest characterisation and not the one a 25.6x day invites. Some context on how ordinary the range is:
- 56.7 percent of all twenty-session observations have been above 1.0x, meaning the funds more than consumed new issuance
- 41.7 percent have been above 3.0x, so today's reading is comfortably inside normal rather than near an edge
- 31.8 percent have been below zero, which is redemptions returning coins, not miners out-producing anyone
The longer sixty-session window tells the calmer version of the same story. It now reads 1.87x — 50,585 BTC against 27,000 mined — up sharply from 0.66x last Tuesday, and at about the 53rd percentile of its own history. Last week's piece predicted that rise too, on the same roll-off logic running the other way, as June's heavy redemptions aged out of the back. Both windows have now converged on something close to the middle of their distributions from opposite directions.
What rolls off next, and what it would take to hold 3.22x
The next five sessions to leave the twenty-session window are August 24 to 28, and they carry a net 11,642 BTC between them — 40.1 percent of the current total. Holding net flow at exactly zero from here, the decay runs:
- After 1 session: 2.75x
- After 2 sessions: 2.30x
- After 3 sessions: 1.98x
- After 4 sessions: 1.64x
- After 5 sessions: 1.93x
The step back up at session five is not an error. August 28 was a -2,594 BTC session, and dropping a negative out of the back of a window raises it. That small detail is a reminder that roll-off is not a smooth headwind; it is a schedule of specific days with specific signs, and anyone extrapolating a trailing ratio without reading the back of the window will be wrong at the turns.
To hold 3.22x through that roll-off, the complex has to replace all 11,642 coins inside five sessions — about $1.01B of net inflow at Monday's close. That is a demanding but not absurd bar: Monday alone supplied 11,528 coins, so a single repeat would very nearly do it. Whether one arrives is not something this series can tell you.
Supply absorption currently normalises to 92.0 out of 100 on the composite institutional score and contributes 18.4 points of the 56.1 total, close to a third of the reading, at a configured weight of 0.20. Friday's signal piece covered the breadth problem underneath that score when it stood at 23.8. The score has since recovered substantially, and a meaningful share of that recovery is one component with a known decay schedule in front of it.
How much bitcoin do ETFs buy over a full cycle?
The rolling windows swing between -8.6x and +11.8x. The cumulative total does not swing at all, and it is the more durable expression of what this channel has done to bitcoin scarcity.
Since January 2024, net ETF creations converted at each day's close total 668,193 BTC. Set against the 343,350 BTC the network issued across the same 691 sessions, the life-to-date ratio is 1.95x — for every coin mined since the funds launched, they have taken just under two off the market. At today's 450 a day, the accumulated stack represents 1,485 days of production, a little over four years' worth, and about 3.18 percent of the 21 million coins that will ever exist.
That is the number worth carrying. It compounds through the redemption weeks, and it is the reason a negative month does not undo the structural point: the funds have been a persistent net drain on available supply for 32 months, at roughly twice the rate of bitcoin daily issuance, through a halving that cut that issuance in half along the way.
Strategy is the obvious second source of absorption and it re-entered the market this week, filing on September 21 for 950 BTC at an average $79,670, taking holdings to 846,000 BTC after three weeks of no purchases. That is a fraction of what the ETFs took on the same day, and the Strategy treasury dashboard tracks it properly — Thursday's article owns that subject.
What this metric does not measure
The absorption ratio is a demand proxy with real edges, and quoting it without them is how a good number becomes a bad argument.
- It counts US spot ETF flow only. OTC desks, exchange accumulation, sovereign buyers and non-US products are entirely absent. The true figure for coins absorbed by all buyers is unknowable from this series and is certainly larger.
- A negative reading means redemptions, not that miners out-produced demand. Coins returning to the market through the creation-redemption mechanism are a different event from coins never having been bought.
- Flow is converted at the daily close, not at execution. On a day that moved as much as Monday, the coin count is an approximation of what was actually transacted.
- It says nothing about price. High absorption and falling prices have coexisted repeatedly in this series, and treating the ratio as a mechanical price input is a misreading of what it is.
What it does measure, cleanly, is ETF demand vs bitcoin mined — a collision between a bid that varies daily and a supply schedule that does not. This week that collision produced one of the six largest single sessions on record and a trailing reading that fell anyway. Both are true, and the second is the one to watch into next week.