Coin Metrics Revises 19 Months of Bitcoin ETF Wallet Data

Coin Metrics has corrected 19 months of historical Bitcoin ETF data across 60 network-data series, a broad revision that reaches far beyond a single flow chart. The affected data runs from February 7, 2025 through September 17, 2026 and includes 35 daily and 25 hourly series covering deposits, withdrawals, transfers, transactions, net flows and supply.

The scale of the rewrite is clear. Its meaning is not. Coin Metrics said users can backfill the revised data, but its public disclosure did not set out the cause of the recalculation, the old and new values, an aggregate difference, a percentage change, or even whether the changes were predominantly upward or downward, according to CryptoSlate’s account of the update.

That leaves a gap between the operational fact of a correction and the financial conclusions drawn from the data. These are not official fund-flow records. They are on-chain measures built from Coin Metrics’ identification of ETF-controlled wallet addresses. A historical revision may therefore reflect a better address map or a custody-related reassessment rather than a newly discovered change in ETF investor demand.

The 19-month revision reaches 60 Bitcoin ETF network-data series

Coin Metrics’ September 22 status notice describes a correction to Bitcoin ETF-related historical network data, covering the period beginning February 7, 2025 and ending September 17, 2026. The 60 affected series comprise 35 daily and 25 hourly metrics, according to the Coin Metrics status page.

The affected categories matter because they cover several ways analysts may seek to describe ETF activity on Bitcoin’s ledger. Deposits and withdrawals can be used as a proxy for coins moving into and out of identified ETF wallets. Transfer counts and transaction measures can be used to assess operational activity. Net-flow and supply series, meanwhile, can feed broader work on ETF accumulation, distribution and the share of Bitcoin attributed to these products.

A correction across all of those categories is different from a technical amendment to one calculation. The daily data can shape longer-run comparisons and historical narratives, while hourly series are more likely to be incorporated into event studies, short-term market analysis or timing-based research. Revising both means that a dataset downloaded before the correction may no longer align with the currently available version even if a user’s methodology has not changed.

Backfilling gives users a route to replace prior observations, but it does not reveal what changed inside the database. Because Coin Metrics has not published deltas, a researcher cannot tell from the announcement alone whether an apparent historical pattern was slightly adjusted, materially altered, or left broadly intact despite the correction.

Coin Metrics’ ETF flows are address-attribution metrics, not official fund flows

Coin Metrics’ ETF coverage includes identified addresses associated with BlackRock, ARK Invest, Franklin Templeton, Grayscale, Invesco, Valkyrie, VanEck, WisdomTree and Bitwise, as well as aggregate ETF-flow series, according to its ETF network-data documentation.

Within that address universe, Coin Metrics defines deposits as assets sent to addresses it identifies as ETF-owned and withdrawals as assets leaving them. For Bitcoin, the calculation excludes change outputs to reduce double-counting of coins returned to the same ETF cluster, as described in the firm’s methodology documentation.

The series consequently offers a structured view of attributed on-chain wallet movements. Its reading depends on which addresses are included: a blockchain transaction does not by itself say whether the movement was a share creation, a share redemption, an internal custody operation or a newly associated address.

CryptoSlate reported that the revised wallet metrics are not equivalent to official ETF creations, redemptions or fund-accounting records. The historical correction should therefore not be described as a restatement of official ETF flows. Wallet-cluster reclassification or custody changes can alter the series without showing that ETF investors bought or sold shares.

Backfillable data without public deltas limits historical ETF-flow analysis

Coin Metrics made revised Bitcoin ETF-related historical data available for backfill, but its public disclosure did not provide before-and-after values, aggregate BTC changes, percentage adjustments or the general direction of the revisions. The central limitation is therefore not the correction itself, but the inability to measure the historical break from the announcement alone.

Users can replace affected observations if they have access to the backfill, but they cannot fully reassess conclusions drawn from the prior series using the public disclosure alone. The affected daily and hourly data cover a 19-month period and include flow and supply metrics, making data vintage relevant when earlier analysis is updated.

These are attributed on-chain wallet movements, not official ETF creations, redemptions or fund-accounting records. Coin Metrics’ series depend on identified ETF addresses, so changes in address attribution or custody classification can alter the historical record. Interpretations of rising supply as absorption or withdrawals as distribution may consequently need to be retested, without implying that every prior conclusion is wrong.

Coin Metrics provides issuer-specific and aggregate ETF series, allowing comparisons across products. But the public disclosure does not show whether the revisions were concentrated in a particular issuer, metric, frequency or part of the covered period. For prior work, the useful update is a comparison of old and revised outputs rather than a refreshed chart alone.

The stakes rise with ETF holdings near 1.6 million BTC

The importance of precision grows with the scale of the market being measured. A 2026 presentation filed with the U.S. Securities and Exchange Commission cited Dune and ETFGI data showing that U.S.-listed spot Bitcoin ETFs held approximately 1.6 million BTC by the first quarter of 2026. The figure is a scale comparison rather than a measurement of Coin Metrics’ revised series, but it shows why ETF supply and flow proxies have become material inputs to Bitcoin market-structure analysis.

At that scale, an attributed-wallet dataset can influence how observers frame questions about available supply, custody concentration and the relationship between ETF activity and Bitcoin’s broader market. A supply metric is not merely descriptive when it is used as an input to such arguments. Its address coverage and revision history become part of the analysis.

Still, the SEC-filed presentation’s cited holding estimate and Coin Metrics’ network metrics should not be treated as interchangeable. The former provides a reported market-scale reference drawn from Dune and ETFGI data; the latter is a suite of on-chain series based on identified addresses. Their proximity in subject matter does not make them equivalent records, and the 1.6 million BTC figure does not reveal the size of Coin Metrics’ historical adjustments.

That unknown numerical impact is the unresolved feature of the episode. Coin Metrics has identified the period and breadth of the correction and made the revised data available for backfill. Until a public explanation provides the cause and size of the changes, however, the revision is best understood as a disclosure and measurement issue—not verified evidence of a 19-month change in Bitcoin ETF investor flows.

Disclaimer: This article is provided for informational purposes only. It is not offered or intended to be used as legal, tax, investment, financial, or other advice.