Bitcoin (BTC) ETFs Saw Outflows in Seven of Eight Selloffs as Declines Slowed
A review of 694 trading days found that ETF flows tracked Bitcoin’s recent moves more closely than its returns over the next five trading days.

U.S. spot Bitcoin (BTC) exchange-traded funds posted net outflows in seven of eight selloffs of at least 15% since 2024, while their flows tracked recent prices more closely than near-term returns.
The review covered 694 trading days through Oct. 16, 2026. It defined a selloff as a decline from a peak daily close to a low daily close. ETF flows were measured at the 4 p.m. ET close, while Bitcoin prices used daily closes at midnight UTC.
The exception came from May 20 to July 7, 2024, when Bitcoin fell 21.8% and the funds recorded $2.190 billion in net inflows. The largest withdrawal period ran from May 10 to June 30, 2026: outflows totaled $8.205 billion during a 28.8% decline, and funds posted outflows on 88% of trading days. Two earlier periods also saw large withdrawals: $3.386 billion from January through April 2025 and $6.922 billion from Oct. 6, 2025, through Feb. 5, 2026.
Selloffs moved more slowly during the ETF era, but their median depth was not consistently smaller. The median decline was 21.3%, at a daily pace of 0.69%. In 2022-23, the median was 19.4% at 1% a day; in 2021, it was 24% at 1.37% a day. Annualized volatility was 47% in the ETF era, compared with 55% in 2022-23 and 81% in 2021. Selloffs of at least 15% also occurred about 2.9 times a year, down from 4.4 times in 2022-23 and eight times in 2021.
ETF flows had a 0.54 correlation with Bitcoin returns over the prior five trading days, compared with 0.03 for returns over the next five. The totals include Grayscale’s GBTC, so transfers from that fund into other spot ETFs are not counted as new buying across the group.
The comparisons do not establish that ETFs slowed selloffs. Bitcoin volatility was already declining before the funds launched, and eight selloffs are too few to prove a statistical relationship. DWF Labs Research summarized its finding: “ETF flows confirm moves; they don’t predict them.”