On October 22, 2026, US spot Bitcoin (BTC) ETFs recorded $635.6 million of net outflows, according to Farside Investors. That figure measures a specific fund universe and session. Comparing it with CoinShares, Glassnode, Kaiko or CoinGlass requires matching the instrument, measurement period and methodology before interpreting any difference.
What happened
Our report on the October 22 sell-off used ETF flows alongside derivatives information. Rechecking Farside on October 25, 2026 confirms the historical outflow. Its table identifies individual funds, dates and US-dollar millions, and warns that the automatically generated data can contain errors. Those labels belong with the number whenever it is reused. Farside's table is evidence about those funds, not every transaction involving Bitcoin.
Where analysts disagree
Here the distinction is methodological, rather than a dispute over a forecast. CoinShares research author James Butterfill's March 23, 2026 fund-flows report separates weekly, month-to-date and year-to-date flows from assets under management. Its tables span providers, assets and countries and cite Bloomberg and CoinShares. CoinShares itself appears among the providers, making its commercial involvement visible. This dated report illustrates the reporting structure; its March figures are not current market readings.
Glassnode's exchange-data transparency notice addresses a different problem: identifying wallets belonging to exchanges. It combines verified addresses, external labels and clustering techniques. The analytics provider acknowledges incomplete coverage, retrospective revisions and internal wallet reshuffling. Its position is that address classification needs explicit limitations, not that a blockchain transfer automatically proves selling.
Kaiko's April 13, 2026 rates rulebook, defines reference prices derived from executed trades on centralized spot exchanges, with exchange selection and scheduled reviews. Kaiko supplies the benchmark methodology; it is not presenting an independent judgment about another vendor's accuracy. These documents support different measurements, not a ranking of providers.
Why it matters
Confusing these categories can turn a measurement difference into an unsupported market narrative. A fund-flow series and a wallet-balance series do not identify the same activity. Glassnode explicitly excludes exchange liabilities from this balance dataset, so the resulting asset count cannot establish solvency. Its warning about internal transfers also limits claims that a large movement represents fresh selling. Glassnode's methodology supplies those boundaries.
Similarly, CoinShares labels assets under management separately from flows. Treating the former as money entering during the week would contradict the report's table definitions. Our ETF-flow explainer addresses the separate question of how fund activity relates to price. The editorial implication here is narrower: establish what was measured before building a causal argument.
How should you compare crypto data providers?
Use the individual dataset as the unit of comparison. A provider can publish several kinds of information.
| Provider | Dataset examined | Main comparison check |
|---|---|---|
| Farside | Daily Bitcoin ETF flows | Fund universe and US$ millions |
| CoinShares | Investment-product flows | Period, geography and asset |
| Glassnode | Exchange wallet balances | Labels, coverage and revisions |
| Kaiko | Reference rates | Selected trades and methodology |
| CoinGlass | Market-data API | Exact endpoint and market type |
The table summarizes the linked documents, not each firm's entire catalog. CoinGlass's API introduction lists derivatives, spot, ETF and on-chain datasets, including open interest, funding and liquidations. It does not establish that every endpoint shares one calculation method. A comparison needs the selected endpoint's definitions; the overview alone cannot certify completeness or reconcile competing liquidation totals.
Our short-squeeze explainer provides related context for interpreting forced-position closures. Its historical figures are not being updated here.
What to watch
For the next comparison, preserve the retrieval time, measurement window, currency, instrument list and revision policy. Check whether a missing entry is actually zero before including it in a total. Farside's dated rows provide the fund-level structure; Glassnode's notice explains why address-based history can change. For Kaiko prices, record the applicable rulebook version rather than assuming every benchmark uses an identical trade universe.
Kaiko's resources page also lists an October 22, 2026 announcement about updated benchmark names and identifiers, and a September 21 exchange-list update. An older rulebook is useful for understanding the framework, but does not establish the current constituents or identifiers of a particular benchmark.
Bottom line
- Farside reports $635.6 million of net Bitcoin ETF outflows for October 22, 2026. Source
- CoinShares separates fund flows from assets under management. Source
- Glassnode warns that exchange-address data can be incomplete and revised. Source
- Kaiko reference rates and CoinGlass's multiple API datasets require dataset-specific interpretation. Kaiko; CoinGlass