Bitcoin ETF flows track recent price moves more closely than subsequent ones, according to DWF Labs research published October 22. The findings challenge a familiar shortcut: treating a large daily inflow as a forecast.
TL;DR: DWF finds Bitcoin ETF flows track past returns more closely than future returns.
The study examined 694 trading days through October 16, using aggregate U.S. spot ETF flows from Farside and Coinbase BTC-USD prices. It reported a correlation of 0.54 with the preceding five-day return, compared with 0.03 for the following five days.
The distinction is between observing demand and forecasting returns. Two series moving together does not show which caused the movement. A relationship with earlier prices may be useful for interpreting investor behavior without supplying a reliable rule for what happens next.
An ETF Purchase Can Have A Hedge Attached
Interpreting the economic motive behind a flow requires more than its direction. CME Group’s explanation of crypto basis trading describes how an investor can buy a spot ETF and sell futures at the same time when futures trade at a premium.
The position aims to capture the difference between the two prices as they converge. Its combined exposure can therefore be substantially different from buying Bitcoin outright in anticipation of a higher price. Financing, trading costs and the terms of the hedge affect the outcome.
CME also notes that several large spot Bitcoin ETFs and its futures use related reference-rate benchmarks. That alignment helps institutions pair the two sides. An observed ETF purchase can consequently belong to a trade whose full risk position is invisible in the fund-flow total.
Creation Data Is Only Part Of The Picture
The plumbing changed during the study period. In July 2025, the SEC permitted in-kind creations and redemptions for crypto exchange-traded products by authorized participants. That allows qualifying transactions to use the underlying assets instead of requiring cash. It does not give every retail shareholder a direct redemption right.
These mechanics make fund flows useful evidence of changes inside an exchange-traded wrapper, while leaving questions about investor intent and transactions elsewhere. Holdings under fund custody reveal only one part of a participant’s total position.
A near-zero forward correlation should be read narrowly. It does not prove that ETF activity can never move Bitcoin’s price, or rule out effects within a trading day. Daily observations cannot, by themselves, settle the sequence of every purchase and price move inside that day.
The practical implication is a higher bar for predictive claims. A useful forecast needs evidence that it works on subsequent observations and survives costs and changing market conditions. A large flow number, even when accurately measured, is only the starting observation for that test.
Written by the News Desk and edited by Samuel Rae.
