ETF Flow Focus · Published

Bitcoin ETF Inflows: +$385M Over 60 Days, and It's All IBIT

Bitcoin ETF inflows just crossed a line they had not crossed since May

The trailing 60-session net flow for US spot bitcoin ETFs turned positive on August 27, ending a 62-session stretch below zero, and closed the week at +$385.0M. That is the category's three-month demand picture flipping from negative to positive for the first time since May 28. It is also, on its own, the thinnest possible version of good news: BlackRock's IBIT contributed +$729.3M over those same 60 sessions, which means every other product in the category combined ran -$344.3M.

Both facts are true at once, and the gap between them is the most useful thing in the data as this week opens. Last week's spot bitcoin ETF flows were positive, the month was the best of 2026, and the three-month trend has recovered — and none of it describes broad demand.

Every figure below is read from bryptoflows' own flow dataset, sourced from Farside Investors and current through August 28. All 12 tracked products reported.

Why the three-month trend flipped, and how much of it was arithmetic

Last Monday's flow analysis set a falsifiable bar: the 60-session window would turn positive if and only if the week ending August 28 delivered more than about +$539.5M, because roughly $1,748.1M of negative flow from late May and early June was scheduled to roll out of the window regardless of what new money did. The week delivered +$924.5M, and the trailing 60-session figure moved from -$2,287.6M to +$385.0M in five sessions.

The split is worth stating precisely, because the headline reads better than the mechanics. Of that roughly $2.67B improvement, about $1.75B was the calendar — old outflows aging out of a rolling window — and about $0.92B was new money. The trend turned because June stopped counting, not only because August bought. That does not make the crossing meaningless: a window that has shed its worst stretch is a different starting point than one still carrying it. But anyone reading "three-month bitcoin ETF flows turn positive" as evidence of accelerating demand is reading about two-thirds arithmetic.

Strip out BlackRock and the recovery disappears

The ex-IBIT number is where the issuer split earns its keep. Over the trailing 60 sessions the category is +$385.0M and IBIT alone is +$729.3M, leaving the other eleven products at -$344.3M. That ex-IBIT measure has been negative for 73 consecutive sessions, last positive on May 14 — it did not cross zero last week, and it is not close to crossing.

One honest qualifier, because it cuts against the alarming reading: this is the category's normal state, not a new fragility. Of the 445 sessions in which the trailing 60-session flow was positive, 261 — 58.7% — had the ex-IBIT figure negative at the same time. IBIT carrying the complex is not this month's development; it is the structure of the category.

What is specific now is the degree. IBIT took 101.5% of last week's net flow and 60.2% of the $1,669.3M of gross activity across all 12 products. Life-to-date its +$63,364.7M is a 115.8% contribution to a category total of +$54,699.6M — the fund has taken in more than the entire category has, because Grayscale's -$24,769.9M works the other way.

Track the live issuer split on the ETF flow dashboard and the category pace on the flow pulse overview.

The Grayscale trade nobody is describing: GBTC out, mini trust in

Grayscale is usually written up as one line — the category's structural drag — and at the issuer level that is fair. But its two vehicles have moved in opposite directions long enough that one number hides the actual trade. Over the trailing 60 sessions, legacy GBTC shed -$902.8M while Grayscale's newer BTC "mini" trust took in +$586.8M; over 120 sessions it is -$1,696.4M against +$692.0M. The pattern holds month by month across 2026:

Life-to-date, GBTC sits at -$27,605.3M against the mini trust's +$2,835.4M, on 64 positive sessions out of 660 reported days versus the mini trust's 210 out of 521.

The flow data cannot tell you these are the same dollars. It shows two funds at the same issuer, one bleeding and one gathering, over the same period, in a rough ratio of three out to one in — consistent with holders rotating from an expensive legacy vehicle into a cheaper sibling rather than leaving bitcoin, but "consistent with" is as far as the dataset goes. What matters for the week ahead is directional: about a third of GBTC's outflow is staying inside the category, so reading GBTC alone makes bitcoin ETF outflows look worse than the issuer's net position is.

Morgan Stanley's MSBT has had one negative day in 100

The category's most consistent flow profile belongs to a fund almost nobody writes about. Morgan Stanley's MSBT has reported 100 trading days since first appearing in the dataset on April 8, 2026, and has recorded exactly one negative session: -$5.3M on May 29.

Cumulative flow since launch is +$521.0M, positive in every month it has existed — +$194.4M in April, +$68.9M in May, +$101.5M in June, +$73.5M in July, +$82.7M in August. At +$9.3M it was the only one of the 12 products to post an inflow on August 28, closing a six-session positive streak.

At under 1% of the $54.70B cumulative total, MSBT does not move the category — the consistency, not the size, is the point. Most products here are lumpy, gathering on breakouts and redeeming on drawdowns, which is what makes aggregate bitcoin ETF net flow so reactive to price. A fund that adds a little almost every session describes a different kind of demand, though the dataset confirms the pattern, not the mechanism.

What has to happen this week to keep the trend positive

The roll-off mechanics point in two directions this week, and both are worth stating as numbers rather than sentiment.

The 60-session window is now well protected. The five sessions scheduled to age out — June 4 through June 10 — sum to -$705.2M, so removing them lifts the trailing 60-session figure from +$385.0M to roughly +$1,090.2M before any new flow arrives. For the three-month trend to fall back below zero, this week's net flow would have to be worse than -$1,090.2M. Only 12 of 138 weeks on record (8.7%) have been that bad, six of them in 2026's May–June redemption stretch. Barring a severe week, the three-month recovery is locked in.

The 20-session window works the opposite way, and this is the number to watch. The five sessions rolling out of it — August 3 through August 7 — were all positive and sum to +$865.3M, dropping the one-month base from +$3,322.4M to +$2,457.1M before new flow. The one-month pace therefore declines mechanically unless the category clears +$865.3M, a bar only 38 of 138 weeks (27.5%) have met, and just 4 of 34 weeks in 2026.

That is the real tension as the week opens. The three-month trend improves almost no matter what happens; the one-month pace deteriorates unless this week matches the last two. Both will be true on Friday, and which gets quoted will depend on who is writing. For scale, last week's +$924.5M ranks 37th of 138 weeks — strong, not exceptional, and enough to clear the 20-session bar by only about $59M.

Does one narrow session actually predict anything?

The week ended on a -$201.9M outflow with just 1 of 12 products positive — an 8.3% breadth reading, covered in Saturday's weekly recap as the end of a nine-session run. Whether it carries information is a separate question, and two base rates disagree.

The streak ending says almost nothing by itself. There have been 25 runs of six or more consecutive positive sessions since launch; of the 24 with five sessions of subsequent data, the following five were net positive 14 times — 58.3%, against a 62.0% baseline across all five-session windows. A long run ending is close to noise.

The breadth collapse is a different matter. There have been 111 sessions with breadth at or below 8.33% — 14 in 2024, 41 in 2025, 56 in 2026, so the sample is not confined to one regime. Of the 110 with forward data, the next five sessions were net positive only 42 times — 38.2%, median -$202.1M. The next single session was positive 30.9% of the time against a 58.1% baseline.

Two caveats keep this from being a prediction. These windows overlap heavily, so 110 observations are not 110 independent trials, and narrow sessions cluster inside drawdowns, which does some of the work. And the recent counterexamples cut against it: the narrow sessions of August 10, 11, 12 and 14 were followed by five-session totals of +$56.9M, +$238.4M, +$816.7M and +$1,917.8M. The signal is real in aggregate and has been wrong four times this month.

bryptoflows' composite institutional support signal registers the same tension as one number, and it has moved sharply: 34.2, still "moderate accumulation," but down from 49.0 on August 26 and 51.8 a week earlier, with component agreement of just 35.6%. Supply absorption scores 99.5 and flow momentum 69.9, while flow breadth scores -83.3. Nearly the entire week-over-week decline traces to breadth, not to the size of the flows.

What the flow data cannot tell you

Three limits matter, because the numbers above are precise in a way that invites over-reading. Creations are not buyers. This dataset records creations and redemptions, not intent: a directional allocation and a creation hedged with short futures are identical here. That matters most for IBIT, because The Crypto Times reported that BlackRock has processed roughly $5 billion of tax-deferred coin-to-ETF transfers — existing holders moving spot bitcoin into the wrapper, which registers as inflow without new demand behind it. That is outside reporting, not bryptoflows' data, and the effect cannot be isolated here — but it is reason to treat IBIT's category share as an upper bound on BlackRock's share of new buying.

Flows do not establish causation with price. Bitcoin traded near $78,116 on August 30 after testing roughly $80,210 on August 27, per CoinStats' market analysis, which attributes the reversal to Federal Reserve Chair Kevin Warsh's Jackson Hole remarks on August 28 — citing inflation at 3.7% over 12 months against the Fed's 2% target, and moving market-implied September rate-hike odds from roughly 35% to 56%. Friday's outflow and Friday's price reversal share a date, and the lag between trade date and published creations means this dataset cannot establish which moved first. That is context for why the pattern might break, not an explanation of why it did.

And a single session can undo a trend call. Last Monday's piece identified Fidelity's FBTC as the clearest genuine turn in the data after its one-month flow crossed positive on August 17, and that 20-session figure is still +$318.5M. But FBTC posted -$83.6M on August 27, erasing more than half the fund's five-session total, and its 60-session flow remains -$219.9M. The turn is intact on the one-month measure and unproven on the three-month one.

The summary that survives all three caveats: the three-month trend in bitcoin ETF inflows is positive for the first time since May, mostly by arithmetic and entirely on BlackRock; the one-month pace faces a bar this week that roughly three-quarters of weeks on record have failed to clear; and the week opens off the narrowest breadth reading the category produces. For how fast this reorganises, the July 13 flow review covered a week when the category shed $493M, breadth narrowed to two funds and IBIT went red — seven weeks before the best fortnight of the year.

Flow figures throughout are read from bryptoflows' own API, sourced from Farside Investors and current through August 28, 2026. Price, Federal Reserve and coin-transfer figures are sourced to the reporting linked above, retrieved August 31, 2026.