Institutional Signal · Published

Institutional Bitcoin Demand Doubles to 49.7 in One Week

Are institutions buying bitcoin right now?

Yes, and the pace picked up sharply this week. bryptoflows' composite institutional bitcoin demand score reads +49.7 on a -100 to +100 scale as of the September 24 close, up from +23.8 last Friday. That more than doubles the reading in a week and ends three consecutive weekly declines. The model classifies it as moderate accumulation, and it is built on all eight components, with nothing missing and no weights renormalised — so it is directly comparable to the +23.8, +28.9 and +55.2 readings of the last three Fridays.

The number matters less than what moved it. Last Friday's signal piece showed that 93% of the score's support sat in five sessions about to expire, and that on flat flows the composite would cross zero and land near -15. Flows did not go flat. The five sessions that replaced that block brought in $2,684.3M, and every component but one improved.

What lifted the institutional bitcoin demand score 25.9 points

The composite normalises eight components to a -100..100 range, weights them, and sums the results. Here is the full breakdown in points of the final 49.7, as of the September 24 close, with last Friday's contribution in brackets:

Two components remain net negative, and they are the same two that have dragged for a month: breadth and concentration. But their combined drag has fallen from -14.9 points to -4.1. The model's agreement figure — which measures how much the eight components corroborate each other, not a confidence interval around the score — rose to 56.7 from 43.7. This time, unlike last week, rising agreement accompanies a rising score, which is the combination that actually means something.

The roll-off that was supposed to sink this score never happened

This is the week's real finding, and the arithmetic is worth doing in the open because it is checkable.

Last Friday the composite's 20-session window ran August 20 to September 17 and held $1,935.4M of net flow. The five oldest sessions in it carried $1,797.9M of that — 92.9% of the momentum reading, five sessions from expiry. Those five sessions have now rolled off. On flat flows the window would hold $137.5M, momentum would read about +3.6, absorption would fall to 0.10x, and the composite would sit near -15.

Instead the window now runs August 27 to September 24 and holds $2,821.8M. The five replacement sessions — September 18, 21, 22, 23 and 24 — delivered $433.0M, $999.0M, $714.7M, $346.9M and $190.7M, or $2,684.3M in total. That is an average of $537M a session against the roughly $360M a session needed just to hold the window level.

So the mechanical decay was real, it arrived on schedule, and incoming demand more than covered it. That distinction matters for reading this indicator honestly: the score did not rise because the model changed or because a stale window flattered it, but because five sessions of genuine inflow replaced five weaker aging ones.

Flow breadth went from the biggest drag to the biggest single gain

Of the 25.9 points the composite added, 9.4 came from flow breadth alone — more than any other component contributed. A week ago one of twelve US spot products printed positive. On September 24 five did, one was negative, and six reported exactly zero.

That is progress, and it is also the component readers most often misread, so the limitation is worth stating plainly. Breadth reads only the latest session, not the window. Its normalisation is twice the positive percentage minus 100, so seven of twelve funds must print positive before it crosses into positive territory. September 24 was a quiet $190.7M day at the tail of the streak; September 21, a $999.0M session, had seven of twelve positive and would have scored +16.7 on the same formula. The composite happens to sample the calmer day because that is the latest one.

The honest reading is that breadth improved materially but remains the composite's weakest link, and that a single-session sample makes it the noisiest of the eight. Anyone tracking whether institutional bitcoin buying is genuinely broadening should watch the run of daily readings on the institutional signal dashboard rather than any one print.

Is institutional bitcoin buying broad yet? IBIT still took 85% of Thursday

Concentration says no, not really. BlackRock's IBIT took $162.6M of the $190.7M that came in on September 24 — 85% of the session. Since January 2024 IBIT has absorbed $65,184.9M against a channel total of $57,481.0M, which is 113% of the net and 57.6% of all gross flow, with Grayscale's GBTC at -$27,841.4M on the other side.

That structural imbalance is why concentration still reads -29.7 and why the component has never printed positive. It did improve from an HHI of 0.788 to 0.678 as more funds participated, but an even split across twelve products would read 0.083. Monday's flow piece made the same point from the opposite direction, when Fidelity took 72% of a session and the channel was negative without BlackRock. A one-fund market and a broadly bid market can print the same headline dollar total, which is exactly why the composite carries breadth and concentration as separate components rather than trusting the net figure alone.

What institutions themselves said this week

External reporting this week corroborates the flow data without adding much to it. Bitwise published its inaugural institutional crypto adoption study on September 23, drawn from interviews with senior investment staff at fifteen large institutions including public pension funds, endowments, sovereign wealth funds and investment consultants. Its most relevant finding for this indicator: none of the institutions interviewed cut its allocation through the roughly 50% drawdown between Q4 2025 and Q2 2026, and several added. See the Bitwise release for the methodology.

Treat that as context, not confirmation. A survey of fifteen institutions describes stated intent among a selected group; the composite measures settled money. When the two disagree, the flow data is the one with a settlement date attached.

On the corporate side, Strategy's most recent 8-K, filed September 21, disclosed an acquisition of 950 BTC at an average of $79,670, lifting holdings to 846,000 BTC. That follows 4,603 BTC acquired on August 31. Both filings are purchases — the model reads direction on every filing, and there have been no disposals in the window. Thursday's treasury piece covers the financing mix and the buyback question in detail, and current figures stay live on the Strategy treasury dashboard.

Only 4.3% of the score's support expires in the next five sessions

Here is the structural change last week's piece could not have anticipated, and the most useful forward-looking number in this article.

The five oldest sessions in the current window — August 27, 28, 31 and September 1, 2 — carry a combined $121.7M, or 4.3% of the window's $2,821.8M. A week ago the comparable figure was 92.9%. On flat flows for the next five sessions, the window falls only to $2,700.1M, and the composite barely moves.

The wider picture matches. The oldest ten sessions hold $577.6M, or 20.5% of the window; the newest ten hold $2,244.2M, or 79.5%. Support has shifted decisively from data about to expire to data that just arrived, and twelve of the twenty sessions are net positive.

That cuts both ways, and the symmetry is the point. The composite is no longer vulnerable to a mechanical unwind, but it no longer has an easy tailwind either. From here the score moves on incoming flow rather than on window arithmetic, which makes the next five sessions genuinely informative in a way that last week's were not.

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, weight coverage was complete, and the model never substitutes zero for a missing input — it omits the component and renormalises the rest. That is the right policy, and it is also why disclosure matters: a score computed on five components is a different object from one computed on eight, and this week's is the latter. Readers comparing Fridays should check that count first.

Three limits are worth restating. The composite reads twelve US spot ETF products and one corporate balance sheet, so it measures one visible channel, not 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 designed to lag and will confirm a turn well after the turn, in both directions. And breadth is a one-session sample carrying 14% of the weight.

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 +49.7 says the measurable institutional channel absorbed materially more supply than it released over the last twenty sessions, on support that is recent rather than expiring — not that bitcoin will rise.

All flow, absorption, treasury and composite figures are read directly from bryptoflows' own API, current through the September 24 close and Strategy's September 21 filing, with the spot price quoted live. The Bitwise study is sourced to the release linked above, retrieved September 25, 2026. Current readings stay live on the overview, ETF flow and institutional signal dashboards.