Bitcoin ETF Flows Correlate More With Past Than Future Returns

U.S. spot Bitcoin (BTC) exchange-traded fund flows were more closely correlated with returns over the prior five trading sessions than with returns over the next five, with little measured relationship to forward performance.

The study window ran from Jan. 11, 2024, to Oct. 16, 2026, spanning 694 trading days. Flow correlation was 0.54 against Bitcoin returns over the preceding five sessions, 0.41 on the same day, 0.09 for the next day and 0.03 over the following five sessions. The lead-lag comparison used the 4 p.m. ET close (20:00 UTC during daylight time and 21:00 UTC during standard time). The aggregate included Grayscale’s GBTC, so transfers from GBTC into lower-fee funds counted as offsetting flows rather than new buying.

Large price swings showed a similar pattern. In eight sell-offs, defined as declines of at least 15% from a peak close to a low close, the funds recorded net outflows in seven. Each of the nine rallies of at least 15% coincided with net inflows. They describe co-movement and do not show that ETF selling caused prices to fall.

When Bitcoin dropped roughly 7% from a 20-day high, average daily net flows shifted from $74 million in inflows over the prior five sessions to $72 million in outflows over the next five. Outflows occurred in 19 of the 24 cases that met this threshold.

Drawdowns were less frequent in the ETF era, but their median depth remained comparable with earlier periods. The median decline was 21.3% during the ETF period, compared with 24% in 2021 and 19.4% from 2022 through 2023. Declines of at least 15% occurred an estimated 2.9 times a year in the ETF period, versus eight times in 2021 and 4.4 times from 2022 through 2023.

The figures do not establish that ETF buying caused the slower or less frequent declines; volatility was already falling before the funds began trading. Aggregate flows also do not distinguish activity by retail investors from institutional investors, and the eight-sell-off count leaves a small sample for comparing the largest drops.