Bitcoin ETFs have bought for eight straight sessions, and each one is smaller

Start with the arithmetic that HashKey's Tim Sun handed over, because it is the only part of this story that converts flow into price. He puts the historical relationship at roughly 0.4% of price movement per $100 million of net inflow, with a correlation coefficient below 0.5 ^[10]. Run the eight-day total through that: $2.8 billion is 28 lots of $100 million, so about 11% of price movement, or rather the more interesting version of the sum is that Wednesday's $232.2 million buys you roughly nine-tenths of a percent ^[2][18][19]. Bitcoin moved 1.1% in a day to $79,520 ^[1]. The flow explains the move only if you accept a coefficient the researcher himself calls weak enough to be useless for next-day prediction ^[11].

So the flows are not a forecast, they are a receipt. And what the receipt says is more concentrated than the headline number suggests: BlackRock's IBIT took $2.02 billion of the $2.8 billion, 72% of it, while Grayscale's GBTC shed $50.4 million on Wednesday alone ^[3][7]. That divergence is the whole product category in miniature, with IBIT's cumulative $63.1 billion and GBTC's cumulative $27.6 billion of outflows netting to $54.7 billion for the group since launch ^[8][9]. Subtract IBIT from the sector and the eight-day streak is $780 million across everyone else, which is $97.5 million a day spread over a dozen-odd funds ^[13].

The shape of the run matters more than its size. Daily inflows peaked at $606.3 million on August 20 and have not exceeded $340 million since, with Wednesday the smallest of the set ^[6][4]. Sun reads sustained inflows beyond five days as evidence of spot demand and of a strengthening consensus that the $60,000 to $70,000 correction marked a temporary bottom ^[12]. Fine. But the decay from $606 million to $232 million is a 62% fall in daily size over six sessions, and the honest reading is that price has moved into a zone Sun places against resistance near $85,000, where trapped positions and profit-taking sit ^[15][14].

This is probably wrong, but the test worth watching is Sun's rather than the price: if bitcoin stalls or dips slightly while inflows continue, new capital is absorbing selling, and the bid is real ^[20]. Thursday's slip back under $80,000 is the first data point, and the flow number for it was not yet published ^[5]. Three ways this plays out. Flows keep arriving into flat prices, which validates the absorption thesis. Flows keep arriving while price runs to $85,000, which means you are paying for the same demand at a worse entry. Or the streak breaks, and the eight days turn out to have been the debasement trade that Bloomberg's Eric Balchunas clocked at a combined $7 billion into gold and bitcoin funds in a week, a rotation of attention rather than a durable allocation ^[16].

Note what the money is not doing. Balchunas frames the week as debasement stealing the spotlight from AI ^[16], and Algoz's Stephen Wundke says weekly ETF inflows of nearly $2 billion matched the prior thirty days combined ^[17]. A month of allocation compressed into a week is a reallocation decision inside portfolios that already own something else, not new savings, and Wundke's own marker is defensive: a hold above $72,000 to make pullbacks look like entries ^[21].