Thursday's $730.9m is 74.1% of the $986.9m the funds netted across the whole week [6], which leaves roughly $256m spread across the other four sessions [14]. The digest reports both figures inside the same week, so the concentration is the digest's own arithmetic, not an inference.
The three-week total works backwards unkindly. Subtract $986.9m from $3.8bn and the first two weeks netted about $2.81bn, or roughly $1.41bn each [7], which means the strongest three-week stretch of 2026 [2] closed with its smallest week, about 30% below the pace that built it [8]. Against a $101.3bn asset base [3], the whole $3.8bn is 3.75% of the book [12] - real money, and not the kind of number that repositions a market on its own.
Cumulative net inflows of $55.6bn sit beneath $101.3bn of net assets, a $45.7bn gap equal to 45.1% of the complex [9]; assets are 1.82 times the money net-paid in [10], which puts a crude marker for the average surviving dollar's entry somewhere near $44,000 against Friday's $80,234 [11]. Crude because it ignores fees and nets redemptions at prices nobody recorded, but the direction holds: the dollar that created shares on Thursday bought the same coin at roughly 1.8 times what the average dollar ahead of it paid.
What the series cannot do is name intent. A creation is a creation whether it comes from a model portfolio adding a percent or from the long leg of a hedged position, and the SoSoValue number [2] counts dollars, not reasons. So three readings fit the same data. Allocation flow that arrives lumpy because investment committees meet on calendars rather than on ticks; a handful of discrete large tickets that happen to clear on one session and can clear back out on one session; or a market-wide bid that has little to do with the wrapper at all, since bitcoin's 2.6% week [5] sat alongside ether up 2.3% and XRP up 3% [13], and bitcoin-specific creations do not explain the other two.
My read is that the flow series is the more auditable of the two, since the price still has not held above the 50-week average that Cointelegraph says would confirm the trend [1], and $3.8bn of net creations at least happened in a form somebody had to fund. But a week hanging 74.1% of its total on one session [6] reads as positioning until it repeats, and the deceleration [8] is the awkward fact for anyone treating this as a new baseline.
Four or five consecutive weeks above $1bn with no single day carrying more than half, and the allocation reading earns it; a drawdown that produces net redemptions of similar size and single-session shape, and this was a trade wearing an allocator's clothes.
