US spot Bitcoin ETFs took $999M on Monday 21 September and another $714.7M on Tuesday 22 September, the fourth straight inflow session, and the four-day run is reported above $2.3bn. The week before, ending 18 September, the whole complex took $6.21M. On its face that one Monday session dwarfs the entire week before it.
I'd hold off on reading that as a wave of new believers until Friday, because there's a dated report coming that can tell the two stories apart.
Quick one on the date first. Several outlets dated the $999M print to "Monday, September 22", but the 21st was the Monday and the 22nd was the Tuesday. Go with their date and you've got one session too many, and the $714.7M that really printed on the 22nd starts to look like a contradiction when it's just the next day's number. It's the second time in three weeks I've seen a daily flow figure re-dated to the day it was published, so I check the weekday on these now.
The other way to get an inflow
Cash-and-carry is the trade where a fund buys spot exposure, here the ETF, and shorts CME Bitcoin futures against it to collect the gap between the two. It's delta-neutral and carries no view on price at all, and in the flows table it looks exactly like someone buying Bitcoin because they want to own Bitcoin.
IOSG's weekly brief argues that roughly half of the variation in ETF flows tracks new short positions opened by leveraged funds, at a correlation of about 0.70. Those are their figures and I haven't seen the underlying study, so I'm treating them as a claim with a name on it. I'd still give it some weight, because a crypto venture investor arguing against the industry's favourite bullish statistic is the opposite of talking their book.
If they're right, the $999M and $714.7M sessions could be leveraged funds putting a basis trade back on, and the comparison with that $6.21M week would be measuring a financing spread more than new demand. I've been reading these prints through eligibility, meaning each rule change decides who is able to buy. A basis trade sits outside that frame, all it needs is a spread wide enough to be worth collecting.
Where the hedge shows up
The short leg of that trade lands in the CFTC's Traders in Financial Futures report, under leveraged funds. Their aggregate Bitcoin futures net short was about 39,877 BTC-equivalent in the week ending Tuesday 8 September, after a rebuild of 1,669 on the week, and about 32,602 in the week ending Tuesday 15 September, a narrowing of 7,275. I have both of those figures from a secondary source, derived rather than read off the CFTC's own table, so they carry that caveat.
The report covering Tuesday 22 September publishes on Friday 25 September, and both big sessions sit inside that reporting week, $1.714bn between them.
What Friday can and can't settle
If the leveraged-fund net short widens materially from the 32,602 base, the carry read gets real support, and a good part of that $1.714bn was probably hedged money with no opinion on where $BTC goes next. If the net short comes in flat or narrower in a week that took that much in creations, I think the arbitrage reading fails on its own instrument, which would mean the money most likely wasn't hedged and the demand reading survives.
It won't settle everything though. One week isn't a trend, and "leveraged funds" lumps together hedge funds, CTAs and other managed money, so it isn't a pure arbitrage proxy (the source concedes that itself). And an ETF flow figure on its own can't score this in either direction, however big it is.
Somebody built a product for that hedge
KalshiEX self-certified a Bitcoin perpetual contract, BTCPERP, under CFTC Regulation 40.2(a) on 2 June 2026. Self-certification means the exchange certifies it complies and lists the contract, and the Commission doesn't sign off, so "CFTC approved" is the wrong way to describe it. The filing names who it's pitched at, including "ETP market makers and authorized participants carrying inventory", and argues a perp "eliminates roll cost and roll-date basis risk". That's exactly the crowd whose hedge shows up in the CFTC data as a futures short.
I haven't got a volume or open interest figure for that contract since it listed, so it tells me the hedge is big enough for someone to design a product around it and nothing yet about whether anyone trades it.
$ETH had its own run too, +$162M on 22 September for a third straight inflow day. The same question applies there, I just don't have a positioning series in front of me to test it against.
Bitcoin made an eight-month high on 21 September with rates pointing the other way, and the flows are the easy explanation people reach for. I'll be reading Friday's leveraged-fund line before I make up my mind on Monday's flow print. Let's see how it comes in.