This website uses cookies

Read our Privacy policy and Terms of use for more information.

THE CROSSOVER
PRO
Members-only edition
You have already had last week’s ETF number. Nobody printed the two things that decide what it is worth: how many separate counts it actually represents, and what bitcoin did over the window it covers.
Some issues make a call. Today's issue hands you the method we use on a number you will see quoted all week, so you can check the next one yourself.
MARKETS · the feature

How we read an ETF flow number

Three write-ups of the same bitcoin ETF figure landed on our desk last week. One said $986.7 million. One said almost $987 million. One said $986.9 million. A spread of two hundred thousand dollars across a billion means one underlying dataset, read by three desks on the same day.

The number is real and it is big. US spot bitcoin ETFs took in more money in the week to 4 September than in any week this year, including $730.9 million on 3 September alone, the largest single day since January. That was the third positive week in a row, about $3.8 billion together, and August as a whole took $3.52 billion, the biggest month since September 2025.

Most people read a figure like that as a demand gauge. Money in, price up, and a record week means the institutions have finally turned up. The reading is reasonable on its face, and it is reasonable for a structural reason. The ETFs are the one part of this market that has to disclose what it bought, in public, every day. Everything else, the exchange flow, the treasury companies, the people who have held since the spring, either reports late or never reports at all. A number that gets published daily became the number everybody watches, and it now gets quoted the way a company's earnings get quoted.

Bitcoin was $77,748 on 28 August, at the start of the week that took the money. It closed that week at $79,668 on 4 September. It closed at $78,929 last night and it is $79,246 this morning. Roughly a billion dollars of the most patient, most regulated money in the market went in, and a fortnight later the price is within two percent of where it started.

The same set of reports gives you two answers depending on which window you pick. Take the month and the money looks like it moved the price. Take the fortnight and it looks like it did nothing at all. Almost nobody quoting the figure tells you which window they took.

Every dollar of that billion was paid to somebody who was selling. Nobody publishes who that was.

Continue with The Crossover Pro

The rest of this story, plus two paid editions a week. $9.99/month at the founding rate. Already a member? Sign in above.

Subscribe to Pro