Bitcoin’s US spot exchange-traded funds have generally followed market momentum rather than cushioning declines, with investors withdrawing money during seven of eight major selloffs since the products launched in January 2024. The finding comes from a DWF Labs report covering 694 trading days through October 16 and reported by CryptoSlate.
The report challenges the idea that institutional access has created a dependable source of buying when Bitcoin prices fall. ETF investors supplied demand during every one of the nine rallies examined, but their withdrawals were common at market bottoms and often became heavier as declines progressed.
Record withdrawals during the latest downturns
The largest ETF outflow occurred during Bitcoin’s 28.8% decline between May and June 2026. Investors withdrew $8.2 billion from the products, according to the report. That followed $6.9 billion in withdrawals during the October 2025-to-February 2026 downturn, when Bitcoin lost nearly half its value.
The two episodes exceeded the $3.4 billion redeemed during the January-April 2025 correction. They therefore represent the largest withdrawals from US spot Bitcoin ETFs since their launch.
The selling was particularly pronounced near market lows. In five of the eight corrections studied, the biggest withdrawals took place during the final five trading days before Bitcoin reached its bottom. During the May-June decline, average daily outflows reached $412 million in the final week, while investors withdrew money on 88% of trading days during the correction.
The pattern was reversed in rising markets. All nine rallies of at least 15% attracted net ETF inflows, and seven recorded their strongest buying during the final five sessions before the market peaked. This suggests that ETF activity has often reinforced an existing trend instead of anticipating a change in direction.
There was one significant exception. During the May-July 2024 correction, Bitcoin fell 21.8% while ETFs still attracted $2.2 billion. The decline occurred alongside selling linked to German government liquidations and repayments to Mt. Gox creditors, allowing ETF demand to absorb part of the additional supply.
ETF flows have limited forecasting value
The DWF Labs analysis found that ETF flows had a 0.54 correlation with Bitcoin’s performance over the previous five trading days, but only a 0.03 correlation with returns during the following five days. The same-day correlation was 0.41. In practical terms, flows tended to reflect recent or current price action more than provide a reliable indication of what prices would do next.
Examples from the study support that conclusion. During Bitcoin’s April 2025 rebound, the cryptocurrency gained 9.7%, but ETFs still experienced average daily outflows of $40 million. Inflows later accelerated to an average of $436 million a day after the recovery was already under way.
Similarly, one of the largest daily inflow totals, about $1.2 billion, occurred around Bitcoin’s October 2025 peak, shortly before a prolonged decline.
The largest ETF does not appear to offer a stronger signal. BlackRock’s iShares Bitcoin Trust, known by its ticker IBIT, accumulated $65.6 billion in net inflows during the study period, yet its flows had only a 0.04 correlation with Bitcoin’s subsequent five-day returns. After its largest inflow sessions, Bitcoin rose over the following 20 days in just 47% of cases.
Large purchases also do not necessarily represent a straightforward bullish bet. Investors may move capital between competing funds, while hedge funds can buy ETF shares and short Bitcoin futures to exploit pricing differences without taking a directional view on Bitcoin.
Slower crashes, but still deep losses
The report described a Bitcoin market that has become less volatile and slower-moving since ETF trading began, even though investors continue to sell during corrections. The median daily pace of a selloff declined to 0.69%, compared with 1% in 2022-2023 and 1.37% in 2021.
Corrections of at least 15% also became less frequent, falling to about 2.9 a year from 4.4 in 2022-2023 and eight in 2021. Annualized volatility declined to 47% from 55% in 2022-2023, while the number of annual daily crashes of roughly 7% fell from six to two.
The downturns remain significant. The median correction during the ETF era was 21.3%, compared with 19.4% in 2022-2023. DWF Labs pointed to deeper liquidity, a broader investor base and access to regulated hedging tools such as CME futures and ETF options as possible factors behind the slower market behavior. It also noted that Bitcoin’s volatility had already been declining before the ETFs launched.
The firm estimated that about one-fifth of ETF flows may have been hedged through CME futures basis trades. Leveraged funds’ net short exposure declined from approximately 116,000 BTC in December 2024 to 29,000 BTC. DWF Labs said unwinding those positions may explain much of the selling during the January-April 2025 correction, but no more than roughly one-quarter of the withdrawals during the two largest later declines. That indicates that a larger portion of recent outflows may have reflected investors reducing Bitcoin exposure directly.
The report also found that once Bitcoin fell 7% below its 20-day high, average daily ETF flows shifted from $74 million of inflows to $72 million of outflows over the next five trading days. Selling intensified in 19 of 24 cases, although flows generally returned to neutral within two weeks.
Current market picture
Moneyiar market data at 23:34 on October 22, 2026, showed Bitcoin at $81,250.85, down 3.57% over 24 hours and 1.04% over seven days. Solana was at $103.23, down 3.50% in 24 hours and 1.95% over seven days. Monad stood at $0.02098, down 4.98% and 8.83%, respectively, while OriginTrail was at $0.3781, down 6.02% over 24 hours and 11.96% over seven days.
Despite the repeated withdrawals during downturns, the 12 Bitcoin ETF products had accumulated $57.96 billion in cumulative net inflows and managed about $110 billion in assets, according to the report. The figures show the scale of institutional participation, but the flow history suggests that participation has not consistently translated into support during market stress.
