Institutional Signal · Published

Institutional Bitcoin Demand Drops to 17.4, Six-Week Low

Are institutions buying bitcoin right now?

Barely, and by a far thinner margin than a week ago. bryptoflows' composite institutional bitcoin demand score reads +17.4 on a -100 to +100 scale as of the October 8 close, down from +42.4 last Friday. That 25.0-point fall drops the model out of moderate accumulation into neutral or mixed — the band it reserves for readings between -20 and +20, where it declines to call a direction at all.

The reading is computed on all eight components: the missing-components list is empty, weight coverage is 1.00, and no weights were renormalised, so it is comparable to the +42.4, +49.7, +23.8, +28.9 and +55.2 readings of the previous five Fridays. At +17.4 this is the lowest of the six.

Why institutional bitcoin demand fell 25 points in a week

The composite normalises eight components to a -100..100 range, multiplies each by its weight and sums them. Here is the full breakdown in points of the final 17.4, as of the October 8 close, with last Friday's contribution in brackets:

The eight deltas sum to -25.0 and reconcile exactly to the headline move. Persistence did the most damage at -7.1 points, then breadth at -4.7, momentum at -4.1, absorption at -3.9, and concentration and price confirmation at -2.9 each. Only capacity and accumulation improved, and together they added 0.6.

The model's agreement figure — which measures how far the eight components corroborate each other, not a confidence interval around the score — fell to 43.3 from 52.7. A score this close to zero on agreement this low is the model saying, accurately, that its inputs no longer point the same way.

One fund bought $4.7 million and five sold

The single most arresting number in this week's data is the gross inflow figure for October 8. Across all twelve US spot bitcoin ETFs, total gross buying for the session was $4.7 million — all of it into Franklin Templeton's EZBC. Gross selling was $248.8 million, for a session net of -$244.1 million, which sits at the 13th percentile of all 704 reported sessions with a z-score of -1.11.

The sellers were Fidelity's FBTC at -$197.1M, ARKB at -$20.3M, Bitwise's BITB at -$17.7M, GBTC at -$8.2M and BlackRock's IBIT at -$5.5M. Six products reported exactly zero.

Two things follow. First, IBIT — which has absorbed $65,710.8M since January 2024 against a channel net of $57,156.6M, more than the category's entire net, as Monday's flow piece detailed — was itself a net seller, if only marginally. Second, flow concentration turned negative for a reason worth stating precisely: the HHI of 0.618 is high because one seller dominated, not because one buyer did. Fidelity alone was 78% of the session's gross flow. The component is direction-blind, so it cannot distinguish a concentrated bid from a concentrated exit, and this week it is reading the latter. Last Friday the same component printed positive because selling had spread out. Neither print should be read as support.

Is flow breadth collapsing, or is the component built to read negative?

This is the part of the composite most likely to be over-read, so it is worth calibrating honestly rather than reporting -83.3 as a crisis.

Breadth is the share of reported products with positive flow, mapped to -100..100. One of twelve gives 8.3%, which lands at -83.3. But the denominator includes every product that reported, and a fund reporting exactly $0.0M counts against breadth identically to a fund that sold. On October 8, only five products actually sold; six simply did not trade. So -83.3 describes a session where one fund bought, not one where eleven sold.

Across all 688 scoreable sessions since January 2024, 22.1% have printed breadth at or below -83.3, and in 2026 the median reading is -63.6. This component is chronically negative in the current data, structurally costing the composite roughly 9 points in a median session. Its -83.3 is worse than its own 2026 norm, but not dramatically so — and it hit -100.0 on both September 30 and October 7, when no product at all was positive.

One caveat belongs on the record. Breadth, persistence and concentration together carry 30% of the composite's weight and are all computed from the single latest session, which can still be revised. Monday's piece noted BlackRock's October 2 figure was then missing; it has since arrived at +$158.2M, revising that session from the +$31.7M then published to +$189.9M. A backfill of that size moves all three components, so treat them as provisional.

Next week's reading should rise before it falls

Here is the most useful forward-looking arithmetic in this article, and it cuts both ways across the next three weeks.

The current window runs September 11 to October 8. Thirteen of its twenty sessions are net positive. Breaking the roll-off into five-session blocks shows two sharply different phases ahead:

So the honest forward read is that the composite will likely tick up over the next week for reasons unrelated to new institutional bitcoin buying, then face a mechanical cliff about a fortnight out that current flows are nowhere near clearing. Anyone quoting a higher number next Friday as evidence of recovering demand will be quoting a base effect. Tuesday's supply piece flagged the same quirk from the absorption side.

These are conditional window projections holding the other six components constant, not forecasts. The reference deviation that scales momentum drifts as the window set changes, and new sessions convert to BTC at whatever price prevails, so read them as direction and magnitude, not point estimates.

A data defect in the treasury capacity component

Integrity note, disclosed because this composite's value rests on it. The capacity component reads Strategy's undrawn financing against its bitcoin position, and this run reports $88.16 billion available against a position marked at $70.05 billion — a ratio of 1.259, normalising to +99.8.

That capacity figure is double-counted. The upstream filing data currently carries two rows per at-the-market programme, and the total is a plain sum across rows. De-duplicated, the five programmes — MSTR common plus the STRC, STRD, STRF and STRK preferreds — hold $44.07 billion, which is the $44.09 billion Thursday's treasury piece reported from the same filing, net of the $15.7 million drawn this period.

The effect on the score is small, because the component's curve is already saturated at these levels: the corrected ratio of 0.629 normalises to +94.7 instead of +99.8, contribution falls from +6.0 to +5.7, and the composite would read about +17.1. The headline conclusion does not change, but anyone comparing capacity week to week should know the input moved for a reporting reason, not a balance-sheet one. Live figures sit on the Strategy treasury dashboard.

Three of the four filings in the accumulation window were acquisitions — 950, 1,665 and 334 BTC — and the fourth disclosed no purchase. There were no disposals, and the model reads direction on every filing.

What would move the bitcoin demand indicator next week

In rough order of how much each could shift the score:

What this composite cannot see

All eight components were available this run. The model never substitutes zero for a missing input — it omits the component and renormalises the remainder — which is why the count matters: a score built on five components is a different object from one built on eight, and readers comparing Fridays should check that first.

Three structural limits bear restating. The composite reads twelve US spot ETF products and one corporate balance sheet, so it measures one visible channel of bitcoin institutional adoption rather than institutional demand in general; direct custody, offshore vehicles and private mandates are invisible to it. Its two heaviest components are 20-session trailing windows, so it is built to lag and will confirm a turn well after the turn, in both directions. And its three single-session components carry 30% of the weight on one day's data the upstream source may still revise.

One methodology note: the price component's window is 20 price observations on a calendar basis, while the flow windows are trading sessions, so the two do not span identical periods.

The composite is a directional proxy for institutional bid strength. It is not a price forecast, there is no mechanical link between the score and future returns, and nothing here is investment advice. A reading of +17.4 says the measurable institutional channel still absorbed more supply than it released over twenty sessions, but on support concentrated in one week of September and with every short-horizon component now negative — not that bitcoin will fall.

All figures are read directly from bryptoflows' own API, current through the October 8 close and Strategy's October 5 filing. Live readings sit on the institutional signal dashboard, with the broader picture on overview and detail on ETF flow.