Between 3 and 7 August, US spot bitcoin ETFs recorded $853.54M of net inflows, the strongest week since mid-April, with BlackRock's IBIT taking $693M — roughly 81 cents of every dollar. Bitcoin traded flat to lower across the same stretch. A practical guide to what a flow print actually measures, what it excludes, and why single-issuer concentration limits what it can tell you.
Between 3 and 7 August, US spot bitcoin ETFs recorded $853.54 million in net inflows — the strongest week since mid-April, and five consecutive sessions without a net outflow. Across the same stretch, bitcoin traded in a band of roughly $64,000 to $65,100 and ended it going nowhere. It opened 11 August at $63,912.50, down 1.4% on the session, back at levels seen a week earlier.
For two years the dominant explanatory model in crypto has been that spot ETF flows set the marginal price. Inflows up, price up. Every Friday flow report read as a leading indicator. Last week is a clean counterexample, and it is worth writing as one — carefully, because a single week does not kill a thesis, and the useful takeaway is not 'flows don't matter' but something more precise: flows and price answer different questions, and the market has been treating them as the same question.
The two facts, side by side, are the whole setup. $853.54 million of net creations across five sessions. A price that finished roughly where it started, slightly lower.
The reflex is to reach for an explanation that preserves the model — selling pressure elsewhere offset the ETF bid, or the flows have not settled yet, or the macro backdrop overwhelmed it. Some of that is true, and the macro backdrop genuinely was hostile: the same week carried a Treasury selloff, 10-year yields above 4.7%, and rising odds of a September Fed hike on the back of an oil shock. Risk assets did not have a good week.
But reaching for an offsetting explanation every time flows and price diverge is how a model survives without ever being tested. A better response is to ask what the flow number was ever capable of telling you.
A spot bitcoin ETF flow figure is a creation-and-redemption number. When demand for shares exceeds supply, authorised participants create new shares and the fund acquires the corresponding bitcoin; when it runs the other way, shares are redeemed and bitcoin is sold. The daily net figure is the difference.
Three things follow from that, and each one limits what the number can support.
Put those together and the week stops looking anomalous. $853 million is a meaningful number in isolation and a small one against global daily turnover. It tells you a particular category of buyer showed up. It does not tell you what everyone else did.
The second observation is the one that should change how you read the series going forward. BlackRock's IBIT accounted for $693 million of the $853.54 million — roughly 81 cents of every dollar that entered the category.
That is a fact about market structure more than about sentiment. When a flow series is dominated by a single product, it is largely a readout on one issuer's distribution channel: which wealth platforms have approved it, which model portfolios include it, and where those platforms are in their allocation cycle. It is a proxy for one firm's institutional footprint, and reading it as 'institutional demand' in the aggregate overstates what it can bear.
The practical implication is about variance. A series driven by one distributor is lumpy by construction — a single platform decision can produce a headline weekly number in either direction, and neither print is telling you much about the underlying breadth of demand. A $850 million week where the money came through eight issuers would be a genuinely different signal from one where it came through one. The headline figure does not distinguish them.
To be fair to the concentration: IBIT's dominance also reflects real advantages — liquidity, spreads, and BlackRock's distribution reach — and investors choosing the most liquid vehicle is rational rather than sinister. The point is not that concentration is bad. It is that it makes the aggregate flow number a narrower instrument than its framing suggests.
One more number belongs in the frame, and it is the one most of the week's coverage omitted: year to date, US spot bitcoin ETFs are still down roughly $4.5 billion in net outflows, despite the recent rebound.
That reframes the story considerably. 'Strongest week since April' is accurate and also a statement about a low base. A five-session inflow streak inside a year of net redemptions is better read as a tentative stabilisation than as institutions piling in. Both descriptions of last week are true; only one of them survives being zoomed out.
This is the general hazard with flow headlines. Weekly figures are published weekly because that is the reporting cadence, not because a week is the right window for the question most readers are asking. A number framed as the best since April is telling you about April as much as about now.
A short framework, useful the next time a flow headline crosses.
And be honest about the limits of the observation this article is built on. One week is a short window. Flows settle with a lag, so the alignment between a creation print and the price action it corresponds to is approximate. A single divergence does not establish that ETF flows never influence price — there is a reasonable case that over longer horizons, sustained net creation removes supply from the market and does matter. What last week does establish is that the relationship is not tight enough to trade on weekly, and anyone who was reading Friday flow reports as a Monday price forecast has now seen why.
The version of this story that would have been easier to write is that institutional demand is back and the price will follow. It might. But the flow number does not say that, and it never did — it says that $693 million came through one distributor in a week when the price did not move. That is a smaller and more useful fact.
Because ETF creations are one channel of demand, not all of it. Price is set at the margin by everyone — offshore spot exchanges, perpetual futures and the leverage on them, miners selling production, corporate treasuries, OTC desks — and no US ETF flow report captures any of that. $853.54M is meaningful in isolation and modest against global daily turnover. The same week also carried a Treasury selloff, 10-year yields above 4.7% and rising Fed hike odds, which was a poor backdrop for risk assets generally.
DrafterDaily covers market structure, flows and regulation with the caveats intact — what the number measures, and what it cannot support.
More crypto analysisA race condition between the display logic and the signing buffer breaks the one promise a hardware wallet exists to make — that what the screen shows is what the key signs — without breaking any cryptography at all.
$26 billion. $31 billion. $38.17 billion. Tokenized Treasuries at $6.8bn, $12.88bn, $15bn, $16.21bn. All published in 2026, all defensible, none reconcilable without knowing the date and the definition.
Coverage fixed on the $75 million exemption. The consequential provision is a conditional safe harbor that would let a crypto asset stop being subject to an investment contract — and a preemption clause almost nobody has read.