ETF Flow Focus · Published

Bitcoin ETF Inflows Go Quiet: $25.2M Across 12 Funds

Where do bitcoin ETF inflows stand as the week opens?

US spot bitcoin ETF inflows enter this week in their weakest short-run position since early summer. The category has posted four consecutive negative sessions totalling -$462.7M, and Friday, September 11 closed at -$13.2M. But the number that actually describes the state of demand is not the net figure — it is the gross. Across all twelve US spot bitcoin ETFs, Friday saw a combined $25.2M of flow in either direction, the sixth-smallest reading in the 669 active trading sessions bryptoflows has recorded since the funds launched in January 2024. The complex did not sell off into the weekend. It very nearly stopped trading.

That distinction is the whole story going into this week, and it is the one the headline outflow number hides. A net reading of -$13.2M is consistent with two very different worlds: heavy two-way churn that happened to settle slightly negative, or near-total dormancy. Friday was emphatically the second.

Friday was the sixth-quietest session on record

bryptoflows tracks every product-level print, so gross activity — total inflow plus the absolute value of total outflow — is directly computable. Friday's $25.2M compares to a 2026 median of $237.7M per session and a trailing 60-session median of $222.8M. Friday ran at roughly one-ninth of a normal day.

Only two sessions in all of 2026 have been quieter: July 27 at $11.6M and April 2 at $15.0M. Against the full history back to January 2024, the only readings below Friday's are November 10, 2025 ($1.2M), May 8, 2024 ($11.5M), the two 2026 sessions already named, and April 11, 2025 ($23.6M). Market holidays, where every fund reports zero, are excluded from that ranking — those are not sessions, they are closures.

The statistical framing supports the same reading. Friday's net print carries a z-score of -0.32 and sits in the 39th percentile of all sessions. For comparison, the category's worst single day on record is -$1,113.7M on February 25, 2025. Friday was not a rout by any measure. It was an absence.

Only three of twelve funds moved at all

Breadth is where the quiet becomes concrete. On Friday, nine of the twelve US spot bitcoin ETFs reported exactly zero. The three that moved:

That is the entire day. The 2026 median is six active funds per session, and the trailing 60-session median is five, so three is genuinely narrow rather than merely unremarkable. Flow concentration over the recent rolling window reads 0.61 on bryptoflows' product Herfindahl index, against a life-to-date product reading of 0.41 — the recent window is materially more concentrated than the category's own norm.

Friday's zeros are reported zeros, not gaps. bryptoflows' data contract holds unreported values as null and never substitutes zero for unknown, and product coverage on Friday was 100% — all twelve funds reported. The funds were open. Nothing happened in them.

BlackRock's bitcoin ETF is still +$2.85B over twenty sessions

Saturday's weekly recap covered where last week's selling was concentrated, so the short version: ARK 21Shares' ARKB at -$234.2M and Grayscale's GBTC at -$129.1M supplied roughly 78% of the week's net outflow between them, while BlackRock's IBIT finished at -$52.5M and is on its third straight negative session.

What is worth adding as the new week opens is the longer window, because it cuts against the natural reading of that IBIT number. Over the trailing twenty sessions, bryptoflows' issuer attribution shows:

A four-session wobble of -$52.5M has not dented a twenty-session contribution of +$2.85B. BlackRock's fund remains the category's engine by a margin that is hard to overstate: life-to-date, IBIT has taken in $64.00B against a whole-category net of $55.22B, a contribution of 115.9%. The arithmetic is not a typo — without IBIT, the US spot complex is net negative life-to-date, largely because Grayscale's GBTC has shed $27.78B over the same span. That structural fact has not changed, and last week did not change it. Whether it is healthy is a separate question, one last Monday's piece examined directly.

Morgan Stanley's MSBT has one negative day in 109

The most under-covered line in the attribution data is Morgan Stanley's MSBT, and it deserves more than the footnote it usually gets.

Since it began reporting on April 8, 2026, MSBT has logged 109 reported sessions. Exactly one has been negative: -$5.3M on May 29, 2026. Sixty-four sessions were positive and the remainder were flat. It is currently on a four-session positive streak, and it was the only fund in the entire complex positive in all four sessions of last week, adding +$19.7M while everything around it redeemed.

The amounts are small — +$559.3M life-to-date, about 0.5% of category gross — so this is not a demand story that moves the aggregate. It is a behavioural one. A fund that almost never prints a redemption looks like steady allocation into a distribution channel rather than tactical positioning, and it is the cleanest example in the data of flow that does not respond to the week's direction. Watch whether that holds through a Fed week; if MSBT prints its second-ever negative day, that would be a more meaningful signal than another -$20M session from IBIT.

$947.8M rolls off the 20-day window this week

Here is the mechanical fact that shapes how this week's numbers will read, regardless of what actually flows.

The trailing twenty-session net currently stands at +$3,310.1M. The four oldest sessions in that window — August 14, 17, 18 and 19 — sum to +$947.8M, and they roll out as this week's four or five sessions roll in. If the category prints exactly zero all week, the twenty-day reading still falls to roughly +$2,362M through pure arithmetic.

Put the other way: bitcoin ETF inflows need to total +$947.8M this week merely to hold the twenty-day window flat. Last week delivered -$462.7M. The week before that delivered +$986.7M, so the required number is not unprecedented — it was cleared in the week of August 31. But it does mean that a merely-okay week will still show up as visible deterioration on the twenty-day view on the ETF flow dashboard, and readers should not mistake that decay for fresh selling.

The nested windows already show the shape of it: -$416.1M over three sessions, -$288.1M over five, +$322.1M over ten, +$3,310.1M over twenty, +$1,605.4M over sixty. Damage is confined to the short end. Month-to-date is still +$307.3M and quarter-to-date +$4,019.2M, though year-to-date remains -$1,370.6M and cumulative net flow sits $7.51B — about 12% — below its all-time peak of $62.74B.

Do four-session outflow streaks usually become five?

The current negative streak stands at four sessions. Across the full history, bryptoflows counts 25 completed negative streaks that reached four sessions. Of those, 14 went on to a fifth — about 56%. Of the fourteen that reached five, nine reached six. The longest completed negative streak on record is thirteen sessions.

So the base rate says a fifth negative session is marginally more likely than not, and that four-session streaks are ordinary rather than alarming: twenty-five prior occurrences in under three years. This is a coin-flip with a slight tilt, not a signal. Anyone presenting the streak itself as evidence of a regime change is reading more into it than the record supports.

What would break the pattern is not subtle. Friday's dormancy means there is very little active selling pressure to overcome — the entire complex moved $25.2M. A single ordinary IBIT session of the kind that was routine three weeks ago (+$200M to +$500M) would end the streak outright and flip the five-session window positive on its own.

Wednesday's Fed decision is the week's real variable

The FOMC meets September 15–16, with the rate decision due Wednesday at 2:00pm ET and an accompanying Summary of Economic Projections and dot plot. That is the only scheduled catalyst of consequence this week, and it lands mid-week, which means Monday and Tuesday flows may well stay in the same low-activity holding pattern Friday established.

Wednesday's macro brief covered the rate-expectation swings running into this meeting. It is worth being explicit about the limits here: bryptoflows can show that flows went quiet ahead of the meeting, but the data cannot establish that the meeting caused the quiet. Positioning ahead of a known event is a reasonable hypothesis and not a demonstrated one.

On the corporate treasury side, Strategy's most recent 8-K, filed September 8, reported no change in bitcoin holdings — neither a purchase nor a sale — with the prior filing on August 31 having recorded an acquisition of 4,603 BTC. Thursday's treasury piece covers that in full; it is Thursday's subject, not this column's.

What the flow data cannot settle

Three honest limits on everything above.

Flow data is not price data, and bryptoflows' candle feed is currently unavailable, so nothing here should be read as a statement about what bitcoin did last week or will do this week. Creations and redemptions also net out authorised-participant activity that may reflect arbitrage, hedging or share-class mechanics rather than end-investor conviction — a redemption is not automatically someone selling bitcoin.

And the quiet itself is genuinely ambiguous. A near-record-low activity session is consistent with buyers and sellers both standing aside ahead of a Fed decision, and equally consistent with the category simply losing the attention it held in August. The flow record cannot distinguish those. bryptoflows' institutional support composite currently reads 26.7, classified as moderate accumulation, but with confidence of only 34.8 — the components disagree with each other, which is exactly what you would expect in a week where most of the complex did nothing. Friday's signal piece works through why that score fell.

The one thing the data does say cleanly: as of Friday's close, US spot bitcoin ETF demand was not being sold down. It was idling. Current readings are always on the overview dashboard.