Bitcoin ETF Inflows Fade in October as Basis Trade Clouds Demand

A big September, then a reversal

US spot bitcoin ETFs went into October with their strongest run in a year. Over the three weeks ending October 2, the funds took in roughly $2.6 billion, even though bitcoin was still about 32% below its October 2025 peak.14 The peak came on September 22, when net inflows hit almost $999 million in a single session. BlackRock's IBIT, ARK 21Shares' ARKB and Fidelity's FBTC accounted for about 91% of that day's total.26 By most counts, September brought about $2.65 billion in net inflows, the second-largest month since October 2025.28

The run did not last. Through October 15, the funds had lost about $532 million on a net basis for the month. Outflows on October 7 and 8 alone came to $729 million, and another $261 million left on October 14, the day a hot September inflation report came out.12 A second tracker puts the month-to-date figure at about $540 million in outflows. It shows total assets across the funds at about $105 billion and cumulative net inflows since the January 2024 launch at about $56.9 billion.11 On-chain research firm Glassnode put it bluntly: the late-September ETF bid faded, and the funds are selling again.33

The headline in this story is still "inflows are soaring." A closer look at who is buying, why, and under what rules points somewhere else. The September surge looks less like a broad return of long-term demand and more like a mix of hedge-fund arbitrage, short covering and one dominant issuer. Meanwhile, US policy has moved from Congress to the regulators.

Who was actually buying

The most useful correction to the inflow narrative is the basis trade. Some hedge funds buy ETF shares and sell bitcoin futures at a higher price, then collect the gap as the two prices converge. The trade pays whether bitcoin goes up or down, so part of any ETF inflow can be arbitrage rather than a bet on the asset.31 CoinShares' James Butterfill said in September that the trade was yielding about 6%. That money leaves once the spread gets too thin to be worth it.3121 One macro research note in mid-October estimated the annualized yield on the CME basis trade at roughly 5.7% to 7.9%, above the 4.19% on two-year Treasuries. Its conclusion was that institutions are treating bitcoin more like a yield product than a directional bet.32

The same reports also show why the basis trade cannot explain everything. Regulatory filings analyzed by CoinShares show hedge funds cut their ETF holdings by 39% in the first quarter of 2026. Professional investors held only 20.8% of US spot ETF assets that quarter, and financial advisers made up 58% of that professional share.21 So the arbitrage money is real but volatile. The steadier base belongs to advisers making long-term allocations.

Off the ETF ledger, on-chain data shows another group of buyers. Santiment reports that wallets holding 100 to 1,000 BTC added about 113,950 BTC between mid-July and late September.21 None of that buying appears in fund flow reports. Most bitcoin changes hands on crypto exchanges and over-the-counter desks, and ETF data misses weekends entirely because the funds trade only when US markets are open.23 Even a near-$1 billion day equals less than 1% of the funds' assets. And that September 21 surge was followed by a week of falling prices.23

Short covering also played a part. Bitcoin futures open interest fell to a five-month low during the late-summer rally, which suggests forced buybacks by short sellers did much of the work, not new long buying.34 Another analysis notes that part of the autumn rebound happened while ETF flows were weak, with flows picking up only later.28

Concentration and conflicting numbers

The fund-by-fund numbers show a split. BlackRock's IBIT is up $261 million so far in October, while the other eleven funds are down about $794 million combined. IBIT holds 62.6% of all the bitcoin in the twelve US spot funds.12 Fidelity's FBTC is the month's biggest loser, with about $461 million in outflows.12 When one fund's inflows exceed the whole category's net total, the overall figure says more about issuer choice than about demand for bitcoin itself.2824

Trackers also disagree on how 2026 stands overall. One puts year-to-date net flows at about $0.4 billion after October's losses.12 Another reported about $1.2 billion before the October outflows.28 A third claimed $5.25 billion in late September, a figure no other source matches.34 Most agree on the shape of the year: about $5.8 billion in net outflows by mid-July, including a 13-day streak from mid-May to early June that removed about $4.4 billion, followed by a recovery in August and September.28

Estimates of the ETF holders' average purchase price also differ. Glassnode puts it near $81,500 and says bitcoin closed above that level after dipping below it on October 8 and 13.33 Bitwise places it closer to $83,000.36 With bitcoin trading around $81,500 to $82,500 in mid-October, the gap matters: under one estimate the average ETF holder is barely in profit, and under the other they are slightly underwater.1230 Either way, Bitwise reports that spot and futures volumes are in the bottom fifth of their one-year range, which makes this a thin market.36

Congress stalls, regulators step in

The policy backdrop changed sharply in September. On September 15, the Senate voted 49–50 on cloture for the motion to proceed with the CLARITY Act, the market-structure bill that would split crypto oversight between the SEC and CFTC. The bill needed 60 votes and failed.23 Republicans Susan Collins, Josh Hawley and Jerry Moran voted against it along with Democrats.3 Some accounts also list Thom Tillis, who switched to no so he could later file a motion to reconsider.75 Democrats focused on ethics provisions they said left out existing crypto ventures and memecoin income connected to officials.3 Prediction-market odds that the bill becomes law in 2026 fell from 82% in February to about 7%. Coinbase stock dropped 8.65% that day, and bitcoin slid toward $76,000.9

Reports of the vote vary on details. One piece dates it to September 2134, and one says the bill fell ten votes short10. The Senate tally of 49–50 against a 60-vote threshold, recorded on September 15, means it fell eleven short.94 Law firm DLA Piper's view is that market-structure legislation is close to dead for this Congress.3

The regulators have moved into the gap. Two days after the failed vote, the SEC and CFTC began building as much of the rulebook as their existing authority allows.5 On October 1, the SEC proposed custody rules that would let registered investment advisers and funds hold crypto through state trust companies, and hold it themselves when no qualified custodian will.41 Public comments are due December 7.48 On October 5, CFTC Chairman Michael Selig announced an advance notice of proposed rulemaking on Regulation CTX and Regulation CAM. These would let exchanges offering retail leveraged crypto trading register with the CFTC.45 Bill sponsor French Hill has argued that agency action is welcome but cannot replace a statute.8

Why the custody rule matters for flows

The custody proposal could change the flow numbers more than the market-structure fight. Advisers hold 58% of the bitcoin-ETF positions reported in 13F filings. Until now, an ETF was effectively their main compliant way to hold the asset.46 Under the proposal, they could also hold crypto directly.46 Our read is that, if the rule is finalized, some adviser demand that now appears as ETF inflows could move into direct custody that the daily flow tables do not track. Persistent outflows would then look worse than actual demand. The proposal's treatment of bitcoin also varies by client type: it is outside the custody rule for advisory clients but inside it for regulated funds, which adds compliance work.43

The takeaway

Read together, the reports suggest that ETF flows are a lagging, partial view of a market driven by arbitrage yields, short-covering cycles, a single dominant issuer and a policy regime now set by agencies rather than by law. The September inflows were real. But a meaningful share probably came from basis trades that can reverse when spreads narrow, and the October outflows show how quickly that can happen. The better signals to watch are sustained flows across several issuers over multiple weeks, whether bitcoin holds the ETF cost basis, and how the SEC finalizes its custody rule.2633 Until then, a strong week of inflows is not good evidence that a new bull market has started.