The number that traveled Friday night was $643.2 million. That is the single-session net outflow across the entire US spot bitcoin ETF complex โ the largest one-day redemption print of the autumn, and the one that flipped October's ledger from green to red. Every desk had it on the tape before the West Coast opened.
But $643.2 million is not the most informative figure in that dataset. The number that matters is 93.2%.
Here is what 93.2% represents. Across the first four sessions of that week โ Monday October 26 through Thursday October 29 โ the whole cohort pulled in $440.8 million. IBIT, BlackRock's fund, took $410.6 million of it. Every other product on the shelf, combined, absorbed $30.2 million. One fund captured ninety-three cents of every dollar that entered the category that week, and then the category went negative anyway.
That ratio never appeared in the reporting I read. The article carrying the $643.2 million headline ran twenty-five paragraphs and mentioned "concentrated demand" once, qualitatively, as an adjective. The single most structural fact in October's flow tape was left unquantified โ and that omission changes what the month means.
Context: the ledger before the ratios
The raw sequence matters, so let me lay it out. The week opened with four consecutive inflow sessions of $110.3 million, $176.0 million, $137.3 million, and $17.2 million. Look at the decay inside that run โ Thursday's print is a rounding error against Monday's. Momentum was already gone before Friday arrived.
Then Friday: $643.2 million out. The distribution was IBIT at $233.7 million, a second large issuer at $201.0 million, a third at $129.5 million. Those three alone were $564.2 million, or 87.72% of the day's damage. Grayscale's GBTC contributed $53.3 million. Another product added $25.7 million. Everything else netted to exactly zero โ not a single offsetting inflow existed anywhere in the complex that day.
Weekly net landed at โ$202.4 million. The month, which stood at +$127.3 million through Thursday's close, settled at โ$515.9 million. Cumulative net inflow since inception remained at $57.05 billion.
I re-derived every identity in that chain before writing a word of this: the 87.72% concentration, the daily decomposition, the weekly subtraction, the IBIT weekly balance of +$176.9 million, the monthly swing. Ten checks, ten passes. The only gap was a $2 million rounding variance on the monthly split โ four cents on the dollar across roughly twenty-one trading sessions. The data discipline here is genuinely good. The arithmetic is not the problem.
Core: the month-end reversal that lasted two trading days
Subtract and October's shape inverts. If the month stood at +$127.3 million at Thursday's close, and the final four sessions alone added $440.8 million, then October 1 through October 23 cumulative was โ$313.5 million. Inside that stretch, October 22 alone was โ$635.6 million, which means the remaining sessions in that window netted roughly +$322 million.
Read that again. The month was in net outflow for three of its four weeks, and its brief "positive" state existed only between the close of Thursday and the close of Friday โ one print, worth 0.22% of the $57.05 billion base. The framing of "October was positive until the last day" is technically true and analytically misleading. It selects the single moment most favorable to a reversal narrative and discards three weeks of steady bleed. Reading the collapse before the narrative breaks means catching exactly this: the collapse had already happened; Friday was where the fiction of stability ended.
The scale problem runs deeper. $643.2 million against $57.05 billion in accumulated stock is 1.13%. The full monthly outflow of $515.9 million is 0.90%. To describe that as demand collapsing is to let a headline do work the arithmetic won't support. Validating the signal amidst the validator noise means admitting that most of what printed in October was noise wearing a signal's clothes.
The real October story was a transfer, not an exit. IBIT's weekly net was +$176.9 million against a market net of โ$202.4 million. That implies every other fund combined shed $379.3 million, and IBIT absorbed 46.6% of that. On the monthly horizon it is starker: IBIT +$932.1 million against โ$1.45 billion elsewhere, an offset ratio of 64.3%. So capital was not leaving bitcoin ETFs in October. It was consolidating into one bitcoin ETF and draining out of the other eleven, and in the final week the drain exceeded the absorber's capacity. The offset ratio slid from 64% to 47% in five sessions. That is a capacity story with a clear trigger date โ and it is a far more actionable read than "outflows hit $643M."
There is a competing explanation the source never entertained, and it is the one I would bet on. When institutions de-risk, they sell what they can sell. IBIT is the deepest, tightest-spread book in the category; it is the instrument you use to raise cash, not necessarily the one you're bearish on. In 2024, while mapping the weekly basis and institutional rebalancing windows for my own ETF work, I watched the same pattern repeat: the largest fund absorbed inflows in calm tape and led outflows in stress tape, purely because authorized participants reach for its liquidity first. A 93% inflow share and a 36% outflow share are not contradictory signals about IBIT โ they are the mechanical signature of the liquidity provider of last resort. The source chose the interpretation that fit its headline. Both readings survive the data.
Then there is the anomaly that got buried. Ethereum ETFs lost roughly $1.61 billion in October โ 3.1 times bitcoin's monthly outflow โ and $431.4 million in that final week, 2.1 times the bitcoin figure. Bitcoin posted one negative session that week. Ethereum posted five. Not a single positive day. That is the most extreme flow signal of the month, and it sat in paragraphs twenty-one and twenty-two, near the end, because the headline needed a conflict structure: BlackRock versus everyone. A full liquidation has no contrast, so it has no narrative value. But readers walking away with "October's problem was a bitcoin ETF month-end redemption" have the direction of the month wrong.
And the denominator problem is fatal to any severity claim: nowhere does the dataset include Ethereum ETF AUM. Without it, $1.61 billion is a number without a scale, and no one can say whether that outflow is a rounding error or an existential bleed.
Contrarian: the fee migration nobody isolated
GBTC printed $53.3 million out on Friday and appears in the redemption table shoulder-to-shoulder with FBTC and ARKB, which invites the reader to file it under the same heading: weakening demand. It almost certainly does not belong there. GBTC carries roughly a 1.5% fee. IBIT carries roughly 0.25%. The steady, months-long drain from Grayscale's book is a price-elastic migration between two wrappers holding the same underlying asset, not a vote against bitcoin. It does not leave the ecosystem โ it arrives somewhere else. The reporting never disaggregates GBTC's monthly share, which is the one unconditionally available number that would separate structural migration from genuine risk appetite.
My Terra-era habit of chasing the alpha through forked trails taught me the discipline: during the May 2022 panic, the honest read came not from the headline outflows but from the cluster of addresses quietly aggregating stablecoins underneath them. Flow direction is cheap. Flow attribution is the product. Here, the attribution was available for free and skipped โ the article concedes, correctly, that flows cannot explain motive, then declines to add even the current day's price action, which would have cost nothing and closed half the gap.
One more blind spot: the entire chain rests on a single aggregator. No issuer's own daily disclosure is cited, no cross-verification exists. It is industry standard practice and not misconduct โ but the arithmetic passing is not the same as the source being confirmed, and a reader should know which one they got. The validator's eye sees what the chart hides, and what this chart hides is that every number above descends from one hand-compiled table.
Takeaway
The number to watch on the next disclosure is not Friday's outflow; it is the IBIT offset ratio. If it climbs back above 60%, October was a liquidity rotation inside a consolidating category and the "reversal" was a scheduling artifact. If it keeps decaying toward 40%, the absorber is full, and the concentration that flattered every headline for six months becomes the fragility that writes the next one. The second number to watch is Ethereum's AUM โ because until someone publishes the denominator, the month's largest outflow will keep being reported in a currency of impressions.
When the logic fails, the chaos begins. October's logic was fine. The framing was not.