Key takeaways
- CoinShares reported US$619M of inflows for the week to 9 March 2026 — not the US$644M that circulated in secondary write-ups.
- Sector AuM fell US$19bn between the 11 May and 1 June 2026 reports while net flows over the same weeks were only -US$4.21bn: roughly US$15bn, or 78% of the AuM move, was price, not investor behaviour.
- Bitcoin's year-to-date net inflow, as reported each week, collapsed from US$4.9bn (11 May) to US$1.2bn (1 June) — 76% of the year's accumulated inflow erased in four reports.
- The 1 June week's US$1,630M of US outflows was 98% of the US$1.67bn global total; the 'global' series is a US series with a rounding error attached.
The series counts ETP tickets, not crypto demand
CoinShares' weekly fund-flow report measures one thing: net subscriptions and redemptions in listed digital-asset investment products — exchange-traded products, trusts and a handful of certificates. It does not measure spot buying. It does not measure on-chain accumulation. When the report for the week to 9 March 2026 recorded US$619M of inflows, with Bitcoin at US$521M and Ethereum at US$88.5M, that was 619 million dollars of net ETP creation. It was not 619 million dollars of net demand for the asset class.
That distinction matters because the wrapper has its own mechanics. The same report logged US$11.4M of inflows into short-Bitcoin products — a flow that is additive to the headline while being directionally opposite to it. A series that adds long and short creations together does not measure conviction.
AuM moved US$19bn while flows moved US$4.21bn
The most common misuse of the series is treating a change in assets under management as a flow. Between the report dated 11 May 2026, when CoinShares put total AuM at US$160bn, and the report dated 1 June 2026, when it had fallen to US$141bn, the sector lost US$19bn of AuM. Net flows across those weeks — US$1.07bn out (18 May), US$1.47bn out (26 May), US$1.67bn out (1 June) — total -US$4.21bn.
The residual, roughly US$15bn, is price. Against a starting base near US$157bn that implies a price effect of about -9% across three weeks, against a flow effect of about -2.7%. Anyone reading the AuM line as evidence of an investor exodus attributed 78% of a market move to a decision nobody made.
Year-to-date is a path, not a position
The chart plots Bitcoin's year-to-date net inflow exactly as CoinShares reported it in four consecutive weeks: US$4.9bn on 11 May, US$3.9bn on 18 May, US$2.6bn on 26 May, US$1.2bn on 1 June. Nothing about January through April changed. The figure fell because it is a running sum, and a running sum with four months of accumulation behind it can be halved twice by three bad weeks.
Reported this way, a year-to-date number is not a measure of positioning. It is a measure of how recently the last outflow happened. A reader who saw US$4.9bn in mid-May and inferred a durable bid was reading an artefact of the summation window.
Persistence carries the signal that single weeks do not
The 11 May report described "a sixth consecutive positive week." The 18 May report described "the first negative week in seven." Both statements are more informative than either week's dollar figure, because they describe a run rather than a print. The 2026 record so far contains one clean six-week positive run and one three-week negative run, with the largest single Bitcoin outflow of the year — US$1,438M — arriving in the third week of the reversal, not the first.
That ordering is the practical point. The largest print came after the trend was already established, which is precisely when a single-week headline is least useful and a persistence count is most useful. A filter that only registers a regime change after four consecutive same-sign weeks would have ignored the 18 May reversal and registered the 1 June week — late, but with the run intact.
The regional split is a US number wearing a global label
On the 1 June report, the United States accounted for US$1,630M of outflows against a global total of US$1.67bn — 98%. Germany contributed US$25.7M, Sweden US$6.6M, Hong Kong US$4.5M. The 16 March report is explicit that "96% of flows originated from the US." The European lines are, week to week, rounding. Treating the series as a read on global institutional sentiment overstates what it can see.
Sizing follows from that. At a 5% digital-asset sleeve in a US$1,000,000 portfolio — US$50,000 — the four-week price effect isolated above (about -9%) is roughly US$4,500, or 0.45% of the total portfolio. The single largest weekly Bitcoin ETP outflow of 2026, US$1,438M, is about 1.0% of the US$141bn of sector AuM it was drawn from. Those are the magnitudes the series can support. It cannot support a claim about what any individual holder did.
What this data cannot tell you
Three limits are worth stating plainly. First, CoinShares' own AuM prints do not reconcile week to week: the 11 May report puts AuM at US$160bn, while the 18 May report describes the prior week as US$159bn. A US$1bn discrepancy in a headline aggregate is a reason to treat AuM as approximate. Second, the reports carry no breakdown of who transacted — a US$1.4bn redemption from an authorised participant unwinding a basis trade and US$1.4bn of retail capitulation look identical in this dataset. Third, the series covers listed products only; if ETP flows and spot flows diverge, this data will not show it, and the premise that flow persistence proxies for demand persistence would break.
The number that would settle it is one CoinShares does not publish: the split between hedged and directional creations. Until it exists, the honest use of this series is narrow — count the runs, discount the prints, and never read AuM as a flow. The shape of the US curve and the ECB's June hike did more to move the discount rate behind these assets in 2026 than any single week in this report did.
This content is for informational purposes only and does not constitute financial advice. Always do your own research or consult a qualified financial adviser before making investment decisions.