How to Read Bitcoin ETF Flow Data Like a Professional: Farside, Creations, Redemptions and the Five Rules

On Monday, July 13, 2026, U.S. spot Bitcoin ETFs recorded $424.7 million of net outflows — the largest single-day exit in nearly three weeks. On Tuesday, Bitcoin rallied 4%. If those two facts feel contradictory, this guide is for you. ETF flow data has become the most-quoted — and most-misread — dataset of the ETF era. Read correctly, it tells you how the largest regulated pool of Bitcoin capital is positioned. Read naively, it produces exactly the wrong conclusions at exactly the wrong moments.

This guide explains what the numbers actually measure, where to find them free, and the five rules professionals use to interpret them — finishing with this week's tape as a worked example.

What a 'flow' actually is (and isn't)

A spot Bitcoin ETF share is created and destroyed by a small club of institutions called Authorized Participants (APs) — market makers like Jane Street, JPMorgan or Virtu. When demand pushes an ETF's price above the value of its bitcoin, APs deliver cash to the issuer, the issuer buys bitcoin, and new shares are created. When selling pushes the price below net asset value, APs redeem shares and the issuer sells bitcoin. A day's 'flow' is simply the net dollar value of those creations and redemptions.

Three consequences follow. First, flows measure primary-market activity only — millions of ETF shares change hands daily with zero flow impact, because buyers and sellers net out on the exchange. Second, flows are a lagging, end-of-day figure: creations settle T+1, so the number you read tonight describes yesterday's imbalance. Third, a zero is ambiguous — it can mean 'no net flow' or 'not reported yet.' Farside's table fills in gradually through the U.S. evening; early readers routinely mistake unreported cells for calm days.

Where to get the data

Four free sources dominate. Farside Investors (farside.co.uk/btc) publishes the reference table — one row per trading day, one column per fund, in US$ millions; it's the source most newsrooms cite. The Block and SoSoValue chart the same data with longer lookbacks and cumulative views. CoinGlass adds holdings-per-fund and premium/discount metrics. Numbers occasionally differ by a few million between sources because of when each snapshots the issuers' files — treat small discrepancies as noise, not news.

The five rules for reading flows like a professional

• Rule 1 — Streaks beat single days. One outflow day is positioning; five consecutive weeks is a regime. The eight-week outflow streak that ran from May into early July 2026 told you institutions were reducing — long before commentary caught up. A single −$425M day, by contrast, often marks event-hedging (as on July 13, the eve of CPI).
• Rule 2 — Decompose by issuer. The complex is not a monolith. BlackRock's IBIT accounts for roughly $60.1 billion of cumulative inflows against $50.9 billion for the entire complex net — meaning everything else, combined, is negative, dominated by Grayscale's GBTC at −$27.3 billion. A "+$90M day" that is 96% IBIT (July 10) is one desk's allocation, not broad demand.
• Rule 3 — Watch for internal shuffles. GBTC charges 1.50%; its own sibling mini fund (ticker BTC) charges 0.15%. When you see GBTC −$53.1M and BTC +$53.4M on the same day (July 13), that is the same money changing seats, not leaving the theater. Net the pair before drawing conclusions.
• Rule 4 — Flows follow price as often as they lead it. Creations spike after rallies because APs arbitrage premium that buyers created; redemptions cluster after breakdowns. Treating flows as a leading indicator inverts cause and effect roughly half the time. Their real value is confirmation: rallies with inflows have sponsorship; rallies without them are running on short-covering.
• Rule 5 — Scale against new supply. Miners issue roughly 450 BTC per day post-2024-halving — about $29 million at $65,000. A +$300M inflow day absorbs ten days of new supply; a −$300M day adds the same in synthetic supply. This is the arithmetic that made 2024's billion-dollar inflow days so explosive and 2026's outflow weeks so heavy.

Worked example: reading the July 2026 tape

Here are the last ten reported sessions from Farside's table, and what a professional reading extracts from them:

The through-line: the early-July recovery was real but narrow, it lost momentum within a week, and the complex entered CPI day defensively positioned. When the soft print hit and Bitcoin jumped 4%, the flow data told you the rally's first leg was happening without institutional participation — which is precisely why this week's July 14–15 flow reports matter more than the price candles they accompany.

The mistakes that cost people money

Three recur. Extrapolating single days — early July's coverage produced multiple claims of long inflow streaks that the actual table contradicts; always count the cells yourself, and beware unreported zeros masquerading as flat days. Ignoring the GBTC drag — cumulative complex figures blend a structural, fee-driven exit that has run since January 2024; strip it out to see marginal demand. Reading flows as retail sentiment — ETF flows are advisor and institutional allocations; retail lives on exchanges and in the Coinbase Premium, a different (and currently negative) dataset entirely.

Daily noise versus weekly signal

A final calibration point: professionals weight the weekly aggregate far more heavily than any daily cell. Daily figures carry settlement noise, reporting lags and single-desk lumpiness; weeks smooth all three. The eight-week outflow streak that defined May and June 2026 was visible and tradable at the weekly level while the daily tape whipsawed between green and red. The same logic now applies in reverse — one strong inflow day after a soft CPI print proves little, but a full positive week across multiple issuers would be the first structural demand signal since spring. If you only have sixty seconds a day for this dataset, spend fifty of them on the weekly totals.

Your 60-second daily checklist

• Open Farside after ~9 p.m. ET, when most issuers have reported.
• Note the total, then immediately decompose: how much is IBIT? Net any GBTC/BTC pair.
• Place today in the streak: extending, breaking, or noise inside a range?
• Scale it: divide by ~$29M (daily mined supply at current prices) for real-economy weight.
• Cross-check price: did flows confirm the move, or did the move happen without them?
• Only then read the headlines about it.