U.S. spot bitcoin and ether exchange-traded product flows are easy to misuse. A single day’s total gets treated as a price forecast, a blank or dashed cell gets treated as a zero, a lifetime total gets treated as a year-to-date figure, and two compilers get added together as if they were the same tape. This guide is a reading method, not a market call. It uses the official public tables at Farside Investors’ Bitcoin ETF Flow page and Ethereum ETF Flow page, as those pages stood when they were retrieved, and it shows how to add, date, and qualify the figures those tables actually display.
The method matters this week because the November 10 rows on both tables are complete, while the November 11 rows are not. A reader who treats an unfinished Veterans Day line as a flow print will mis-state the post-CPI tape. A reader who stops at November 6 will miss the next two completed sessions. The arithmetic below is taken only from completed rows.
What these tables measure, and what they do not
Each Farside page is titled in U.S. dollars millions. The bitcoin table is “Bitcoin ETF Flow (US$m).” The ether table is “Ethereum ETF Flow (US$m).” A cell of 20.0 is therefore twenty million dollars of compiled inflow for that issuer on that day, and a cell of (95.0) is ninety-five million dollars of compiled outflow. The tables are issuer-level daily flow compilations for the listed U.S. spot products, not exchange spot volume, not perpetual-futures volume, and not a net positioning series for the entire crypto market.
They also are not official SEC filings. Farside states, under the table, “The above table is generated automatically. Farside Investors is not liable for any errors or inaccuracies in the data.” The charts carry a second note: “These charts are updated automatically in real time.” Those sentences are part of the source, not fine print to skip. A later revision of a completed day, or a difference versus another compiler, is possible. This guide reports the retrieved official display and tells the reader to re-open the live table before treating any figure as final.
A flow number also is not a return. Bitcoin can fall on an inflow day and rise on an outflow day. Creations and redemptions are one institutional channel; they do not bind the next tick in the perpetual book or in the cash market. The useful habit is to keep three ledgers separate: the flow table, the price, and the derivatives tape.
Units, signs, and the difference between 0.0 and a dash
Parentheses are outflows. A figure without parentheses is an inflow. That convention is consistent across both tables in the retrieved pages. Adding a month-to-date total therefore means adding the signed figures, not adding their absolute values and then guessing the sign at the end.
A completed zero is displayed as 0.0. That is a numeric claim: the compiler recorded no net flow for that issuer on that completed day. A dash is different. On both the bitcoin and ether tables, the November 11 row is a line of dashes in the issuer columns, with a displayed total of 0.0. That row should not be read as a finished session of zero flow. At the time the pages were retrieved, November 11 was still a U.S. trading day — Veterans Day, on which the NYSE is open even though it is a federal holiday — and the tables had not yet filled the issuer cells. This guide leaves that row out of every sum.
The same rule applies whenever a date is on the table but the issuer cells have not been populated. If a weekend or a U.S. market holiday with no cash-session listings produces no row, that is also not a zero flow day in the cash-ETF sense; it is simply a day the table does not cover. Do not back-fill those dates by hand.
Issuer columns, fee rows, and why those fees are not a shopping comparison
The bitcoin table, in header order, shows IBIT, FBTC, BITB, ARKB, BTCO, EZBC, BRRR, HODL, BTCW, MSBT, GBTC, and BTC, plus a Total column. The logos above those tickers, as displayed on the retrieved page, are labeled Blackrock, Fidelity, Bitwise, Ark, Invesco, Franklin, Valkyrie, VanEck, WTree, MS, Grayscale, and Grayscale. The ether table shows ETHA, ETHB, FETH, ETHW, TETH, ETHV, QETH, EZET, MSSE, ETHE, and ETH, plus Total, with logos labeled Blackrock, Blackrock, Fidelity, Bitwise, 21Shares, VanEck, Invesco, Franklin, ms, Grayscale, and Grayscale. This article uses those tickers and logo labels as Farside displays them. It does not invent formal fund legal names beyond that display.
A Fee row sits under the tickers. On the bitcoin table the displayed percentages are 0.25, 0.25, 0.20, 0.21, 0.25, 0.19, 0.25, 0.20, 0.25, 0.14, 1.50, and 0.15. On the ether table they are 0.25, 0.25, 0.25, 0.20, 0.21, 0.20, 0.25, 0.19, 0.14, 2.50, and 0.15. The ether table also has a Staking fee row: N/A, 10%, N/A, N/A, 25%, N/A, N/A, N/A, 5%, 23%, and 6%. Those rows are compiler-displayed fee lines, not a recommendation, not a complete expense prospectus, and not a basis for calling any exchange or any product the cheapest venue. Expense ratios can change, can be waived, and can sit beside other costs. If a reader needs a current official fee, the issuer’s prospectus or factsheet is the document to open, not this blog post and not a flow-table header alone.
The issuer mix inside a daily total is often the real story. A small net total can hide a large outflow in one ticker and a large inflow in another. November 4 on the bitcoin table is the clean example in the current month: IBIT is shown at (201.2) while FBTC is shown at 236.7, and the day’s total is an inflow of 74.6. Quoting only the 74.6 would erase that offset. The same habit applies on the ether side, including on November 10, when a total of (1.3) was the net of a 1.1 inflow in ETHB, a (0.4) outflow in TETH, a (2.0) outflow in ETH, and zeros elsewhere.
Seed figures and conversion asterisks
The ether table includes a Seed row above the first daily date. The displayed seed figures are 10.6, 104.7, 4.4, 2.5, 2.3, 10.2, 1.1, 2.7, a blank in the MSSE column, then 9,199* and 1,023*, summing to 10,360. Farside’s own footnote says, “ETHE & ETH seed is a conversion from a prior product.” Those starred seed amounts are therefore not ordinary first-day creations in the same sense as a later cash-inflow cell. A reader who treats the lifetime Total line as if every dollar in it were a spot-ETF cash creation since 2024 will overstate the cash-inflow history of the two Grayscale ether columns.
The lifetime Total line still includes those seed figures, which is why the footnote exists. When this guide quotes a lifetime Total, it quotes the table’s Total line as displayed and flags the conversion seeds. It does not attempt to “clean” the history by subtracting the seeds unless Farside itself presents an alternative cleaned total, which the retrieved page does not do in the main table.
The bitcoin table, in the retrieved window that begins on October 23, does not show a seed row in the same visible header block. That is a display fact about the retrieved page, not a claim that bitcoin products had no conversions earlier in their lives. Older history lives in Farside’s alternative full-history table, which the page links but which this guide does not re-scrape cell by cell. Lifetime Totals quoted below are the footer Totals on the retrieved main tables.
Footer rows: Total, Average, Maximum, and Minimum
Each table ends with four summary rows. On the bitcoin page, the Total line is 64,810 for IBIT, 10,238 for FBTC, 2,027 for BITB, 941 for ARKB, 140 for BTCO, 303 for EZBC, 329 for BRRR, 985 for HODL, 74 for BTCW, 856 for MSBT, (28,110) for GBTC, 2,879 for BTC, and 55,470 for the grand Total. The Average line’s total cell is 78.0, the Maximum is 1,373.8, and the Minimum is (1,113.7). On the ether page, the Total line is 11,404 for ETHA, 765 for ETHB, 2,434 for FETH, 393 for ETHW, 34 for TETH, 153 for ETHV, 25 for QETH, 65 for EZET, 49 for MSSE, (5,530) for ETHE, 1,974 for ETH, and 11,765 for the grand Total. The Average line’s total cell is 20.3, the Maximum is 1,018.8, and the Minimum is (465.1).
Those Totals are cumulative figures on Farside’s compilation, not calendar year-to-date figures, and not November-to-date figures. GBTC’s lifetime cell of (28,110) and ETHE’s lifetime cell of (5,530) show how a single legacy product can pull a lifetime total well below the sum of the newer tickers. ETHA’s 11,404 of the ether lifetime 11,765 shows concentration in the opposite direction. None of those footer cells should be copied into a sentence that begins with “so far this year” unless the writer has separately added the calendar-year rows.
The Average, Maximum, and Minimum rows describe the history inside Farside’s compilation, not a forecast range. A day can print outside a previously displayed maximum once the table updates. This guide does not treat 1,373.8 or 1,018.8 as ceilings.
How to add a streak or a month-to-date total
A streak is a run of consecutive completed sessions with the same sign in the Total column. It should be dated at both ends, and it should exclude any unfinished row. A month-to-date total is the sum of completed daily Totals in that calendar month, again excluding dashes. Both calculations are arithmetic on displayed cells. They are not official Farside “MTD” or “streak” labels; the retrieved pages do not print those labels as such. When this blog reports a month-to-date figure, it is saying the cells were added by hand from the official daily Totals.
The safe sequence is mechanical. First, confirm the page title and the unit. Second, copy the completed daily Total cells, with signs. Third, add them, then show the work. Fourth, name any day that was excluded because it was unfinished, a weekend, or absent from the table. Fifth, if the paragraph also quotes an issuer, add that issuer’s completed cells separately rather than scaling the total by a guessed share.
That sequence is what the next two sections apply to November 2026, using only completed sessions through November 10.
A worked bitcoin example through November 10
The completed November bitcoin Totals on the retrieved table are (398.6) on November 2, (115.6) on November 3, 74.6 on November 4, (430.2) on November 5, (411.9) on November 6, (201.0) on November 9, and (95.0) on November 10. Adding the signed figures gives 398.6 + 115.6 − 74.6 + 430.2 + 411.9 + 201.0 + 95.0 = 1,577.7 of net outflow, or −$1,577.7 million for the completed month-to-date window. The first five of those sessions, November 2 through November 6, still sum to the −$1,281.7 million week already discussed in this blog’s earlier November outflow recap. The new information after that recap is the completed November 9 and November 10 prints of (201.0) and (95.0).
A four-session outflow streak is visible from November 5 through November 10, which are consecutive completed sessions even though a weekend sits between November 6 and November 9. The signed Totals are (430.2), (411.9), (201.0), and (95.0), which add to −$1,138.1 million. November 4 does not belong in that streak, because its Total is a positive 74.6.
Issuer arithmetic should be shown the same way. IBIT’s completed November cells are (405.3), (95.0), (201.2), (105.0), (403.0), (213.2), and (78.2), which add to −$1,500.9 million. FBTC’s completed November cells are 6.7, 6.8, 236.7, (199.9), 42.8, 0.0, and (36.8), which add to +$56.3 million. The month-to-date bitcoin outflow is therefore not a uniform redemption across issuers. It is concentrated in IBIT, partially offset by FBTC, and, on November 10 specifically, further offset by a 20.0 inflow in MSBT against IBIT’s (78.2) and FBTC’s (36.8).
Those November 10 issuer cells, and the 95.0 total outflow, are the completed session this package’s bitcoin-flow article discusses in more detail. This guide repeats the official cells only so a reader can see the addition. It does not re-tell the market narrative around the already-released October CPI print, and that narrative should not be pasted on top of a table-reading example.
A worked ether example through November 10
The completed November ether Totals are (139.7) on November 2, (85.3) on November 3, 53.0 on November 4, (99.9) on November 5, (93.3) on November 6, (184.6) on November 9, and (1.3) on November 10. Adding the signed figures gives 139.7 + 85.3 − 53.0 + 99.9 + 93.3 + 184.6 + 1.3 = 551.1 of net outflow, or −$551.1 million month-to-date through the last completed session. November 9’s (184.6) is the session already covered in this blog’s ether ETF outflow article. November 10’s (1.3) is a nearly flat completed follow-through, not a second copy of that November 9 story, and not a reason to ignore the month-to-date sum.
The November 2 through November 6 ether Totals add to −$365.2 million. Unlike bitcoin, ether’s November 10 print does not extend a multi-day total-outflow streak in the same way the bitcoin tape does, because the new session is only (1.3) and because the interesting comparison is the collapse in daily magnitude from (184.6) to (1.3). The issuer cells on November 10 are 0.0 for ETHA, 1.1 for ETHB, 0.0 for FETH, 0.0 for ETHW, (0.4) for TETH, zeros through MSSE, and (2.0) for ETH. On November 9, by contrast, ETHA alone was (119.3), with ETHB at (20.8), FETH at (19.1), TETH at (2.0), MSSE at 1.6, and ETH at (25.0). Reading only the two daily Totals would miss that the November 10 near-zero was a broad quieting of the large columns, not a large ETHA redemption hidden by an equally large creation somewhere else.
Concentration still sits in the lifetime footer. ETHA’s 11,404 cell is most of the ether Total of 11,765. A quiet ETHA day is therefore not a minor issuer event, even when the Total is only (1.3). The lifetime ETHE cell of (5,530) remains a warning against treating the 11,765 Total as if every listed product had been a net creator.
Timing, holidays, and unfinished sessions
U.S. spot bitcoin and ether ETPs create and redeem in the cash-market session. A flow table that follows those products will usually complete a day’s row after that session is finished and after the compiler has posted the figures, not at the Asian cash open and not at the moment a CPI number hits the wire. The November 11 dashes are the current illustration. Veterans Day is not a NYSE holiday, so a completed November 11 row may still appear later. Until the issuer cells are numeric, the row is not a print.
Weekends typically do not appear as date rows, which is why November 7 and November 8 are absent and why a “four-day streak” from November 5 to November 10 can be four completed sessions rather than six calendar days. Scheduled U.S. data prints, including the already-released October CPI and the still-unreleased October PPI on Friday, November 13, at 8:30 a.m. Eastern, can move prices on days whose ETF flow rows have not yet been completed, or on days that will later show little net flow. The calendar for trading around CPI, PPI, and FOMC is the place to keep those event times; the flow table is the place to record what creations and redemptions did after the fact.
Because Farside says the table is generated automatically, a prudent reader should also allow for later revisions to a day that already looks complete. If a previously copied cell changes, the later official display, not a blog recap, is the figure to use going forward.
How not to mix compilers, prices, and exchange volumes
Different public compilers can disagree about a day’s total, about when a row is final, and about how conversions or seed amounts are handled. This guide uses Farside because the official pages are public, dated by row, and retrievable. It does not add a second compiler’s number to a Farside number, and it does not average two compilers to produce a “consensus flow.” If another table is introduced in a later article, it should be labeled as that other table, with its own retrieval context noted.
Exchange spot volume and perpetual volume are not substitutes for these cells. A large day in bitcoin perpetual liquidations can coincide with a quiet ETF-flow day, and a large ETF redemption can arrive on a quiet derivatives morning. Adding those series into a single “crypto flow” total is not something the Farside tables authorize. Price is likewise a separate series. The tables do not report closing prices, and this guide does not invent them.
Fee rows, staking-fee rows, and “cheapest product” language should also stay unmixed. The displayed 1.50 percent and 2.50 percent cells on GBTC and ETHE, and the 0.14 percent cells on MSBT and MSSE, are compiler-displayed figures. They are not a ranking of exchanges, and they are not a claim about KCEX or about any perpetual-futures schedule.
A short reading checklist
Before quoting a flow figure, a reader can run a short list. Is the page the bitcoin table or the ether table, and is the unit US$ millions. Is the row a completed numeric session or a line of dashes. Are parentheses being respected as outflows. Is a quoted “total” a daily Total, a hand-added month-to-date, a streak, or the lifetime footer, and has that choice been named. If an issuer is named, were that issuer’s cells added separately. If a seed or a conversion asterisk is in the history, has it been flagged. If the sentence mentions price, volume, or another compiler, have those been kept in separate clauses. If the article is older than the live table, has the live table been re-opened.
Used carefully, that list would have kept November 11 out of any completed-session sum, would have kept the bitcoin lifetime Total of 55,470 from being called a 2026 year-to-date figure, would have shown that IBIT’s November month-to-date of −$1,500.9 million is larger than the market-wide bitcoin month-to-date of −$1,577.7 million would suggest if one assumed equal shares, and would have shown that ether’s November 10 near-zero sits inside a still-negative completed month of −$551.1 million. The tables are most useful when they are read that narrowly, and they are least useful when a single cell is asked to do the work of a forecast.
