Bitcoin’s rally through the Federal Reserve’s September rate hike did not come from a single buyer, and the public record makes that unusually clear. The FOMC lifted its target range by a quarter point to 3.75% to 4.00% on September 16, yet exchange-traded fund data show investors pulled money out of US spot Bitcoin ETFs on the two sessions surrounding the decision before flooding back in. Untangling who actually bought the move that carried Bitcoin to roughly 87,300 USD on September 21 — and back toward 84,000 USD two days later — requires separating three very different groups of purchasers operating on three different clocks.
## The funds first lost money, then made up for it
Farside Investors’ daily fund table records 450.4 million USD of net withdrawals from US spot Bitcoin ETFs on September 15 and another 295.9 million USD on the day of the Fed decision — 746.3 million USD out the door across two sessions. The reversal began on September 17 with 159.5 million USD of inflows, followed by 433.0 million USD on September 18, 999.0 million USD on September 21 and 714.7 million USD on September 22. Completed figures for September 23 added 346.98 million USD for a fifth consecutive positive session, bringing the streak to roughly 2.65 billion USD.
The full picture matters. Subtract the pre-decision withdrawals and the eight trading days show a net 1.9 billion USD entering the funds — a figure that conceals two sharply different periods on either side of the hike.
There is also a data trap for anyone reconciling headlines. An early snapshot of the September 23 table showed only 32.4 million USD, with several issuer rows still blank; the later completed reading included 166.29 million USD for BlackRock’s IBIT and 143.24 million USD for Fidelity’s FBTC. Treating a blank cell as zero would have understated that day’s total by more than 300 million USD.
## Who was buying, fund by fund
Monday’s 999.0 million USD session was broad. IBIT took 381.4 million USD, ARK 21Shares’ ARKB 289.1 million USD and Fidelity’s FBTC 238.8 million USD — about 91% of the day’s total across three products. Tuesday’s 714.7 million USD narrowed to two names, with IBIT at 350.3 million USD and FBTC at 257.4 million USD, roughly 85% of the total.
But a fund vehicle is not a buyer. ETF net flows measure subscriptions and redemptions, not the institutions or individuals placing the underlying orders, and secondary-market share trading can change hands without creating new fund units. A headline can be accurate on flows and misleading on price.
## The corporate buyer with a different clock
Strategy is the one named purchaser in the record. Its September 21 Form 8-K reports 950 BTC bought for 75.7 million USD including fees, at an average of 79,670 USD per coin, between September 14 and September 20, funded from existing USD cash with no shares issued through its at-the-market program. The filing gives a window, not timestamps — it does not show Strategy buying during Monday’s surge. Strive’s September 21 filing discloses another 1,355 BTC at roughly 79,475 USD on average during September 14 to 18. The combined 2,305 BTC is corporate accumulation over overlapping periods; it cannot simply be added to ETF flows and called total demand.
## Some buyers were closing losing bets
The third class did not necessarily want to hold Bitcoin at all. Rising prices force perpetual and futures shorts to buy back exposure, and liquidations can fuel the very move that made the position untenable. Nansen senior research analyst Nicolai Sondergaard described the rally as a combination of ETF demand and short covering, and CoinMarketCap research lead Alice Liu told crypto.news that covering, rather than fresh buying, drove much of the rise. One check: Binance Bitcoin open interest fell from about 5.4 billion USD to 4.9 billion USD between September 21 and 23, though falling open interest alone cannot separate shorts closed in a rally from longs unwound in a retreat.
## Why the divergence on September 23 is the real story
The strongest case for durable demand is the streak itself: five positive sessions spanning both the climb and the pullback, spread across IBIT, FBTC and ARKB, plus two corporate filers. The counterargument is price: Bitcoin did not hold 87,300 USD, and inflows continuing into a falling market mean sellers are meeting the fund cash. Neither reading is wrong — they measure different things.
For a cleaner test, watch whether subscriptions stay large while open interest stops contracting, which would strengthen the case for demand beyond squeezed shorts. If creations dry up while price keeps slipping, Monday’s surge looks more like temporary buying.
What the record can actually name is a set of funds, two companies and a position type — not the owner behind every order. What it rules out are the two simplest stories: Bitcoin did not rally because investors instantly celebrated a hawkish hike, since ETFs shed 746.3 million USD around the decision, and it was not a chart artifact either, with 2.65 billion USD of recorded subscriptions over five sessions. The rally had many buyers, on different clocks.
Market check at publication time: Bitcoin trades near 84,494 USD, Ethereum around 2,685.99 USD, and Solana at approximately 117.05 USD.
retail spot on exchanges shows up in none of these tables. farside covers ETFs, perps show up as OI. the mystery buyer is just everyone using rails nobody indexes
agreed, but even the exchange rail data only shows net flows. wash trading and internal transfers pollute that picture too
746 million out across Sept 15 and 16 and then five straight sessions of inflows after. classic pre-FOMC de-risking. anyone pretending one clean buyer did this needs to read the Farside table again.
three clocks is right. funds trade daily, whales don’t care, spot buyers just DCA through all of it. the record names none of them.
the 999 million day on the 21st is what carried price to 87k. the hike itself was already priced, flows did the rest.
the 999 million day was also momentum chasing, flows follow price both ways. agree the hike itself was priced tho
agreed, and note the 84k pullback came right when inflows cooled to 347 million. the second the bid slows, so does the tape.
746.3 million in ETF outflows right around the hike and btc still ripped to 87,300. the buyers were somewhere else entirely
perps and raw spot on exchanges probably. those inflow days right after the decision were aggressive
the 999M day wasnt one buyer either, it was momentum desks piling into the same trade. people want a mystery whale because its a better story
3.75 to 4 percent and BTC held 84k. tell that to everyone who sold the hike into the close.
^ 84k held through the actual hike, that was the tell. everyone who sold the close got squeezed the next session
The three-clocks framing is right. ETF flow data tells you what wealth advisors did, not what the market actually did.
finally someone says it, fund flow tables get overread constantly