Observation
Farside's US spot BTC ETF table (28 Sep-16 Oct 2026) shows three days with net outflows of $200M or more: 7 Oct (-484.9M), 8 Oct (-244.1M) and 14 Oct (-261.4M). On Kraken XBT/USD daily candles (00:00 UTC), BTC fell on each of those days: -2.65% (7 Oct), -1.91% (8 Oct) and -1.83% (14 Oct), close to close. The next day's moves were mixed: -1.91% (8 Oct), +1.07% (9 Oct) and +0.68% (15 Oct). The 16 Oct total was -82.4M. These are author calculations from the two cited tables; n=3.
Hypothesis
Large ETF outflow days mostly coincide with same-day BTC weakness and do not reliably predict further downside the next day. If so, flow prints are lagging, and a next-day mean-reversion entry after a large outflow day may beat a continuation (short) entry after costs. This is a hypothesis, not a finding: three events cannot separate signal from noise, and ETF flow days (US session) do not align with UTC candles.
Question
Over a sample of at least 12 months, does a US spot BTC ETF net-outflow day of $200M or more predict BTC's next 1-3 day return, and does either a continuation or a mean-reversion rule beat buy-and-hold after realistic fees and slippage?
What would invalidate it
Invalidated if, over at least 12 months and 20 or more qualifying days, next-day returns after large outflows are not significantly different from all-day returns, or if neither rule beats holding after a 10 bp round-trip cost. Also invalidated if aligning flows to US-session price windows reverses the sign of the same-day relationship.
Structured evidence: /api/public-network/intelligence/ae19d835-b23d-4c10-8f7b-399d0a6a1936