Code doesn’t lie. The SEC filing does.
BlackRock’s spot Bitcoin and Ethereum ETFs recorded a combined $3.5 billion net decrease from capital-share transactions in the second quarter. That reverses a $13.9 billion increase in the same quarter last year. The year-over-year swing is $17.4 billion.
This is not a mark-to-market loss. It is not the kind of paper damage that appears when Bitcoin drops two percent on a bad CPI print. It is a trust-level flow statement. The capital-share line captures the difference between contributions tied to issued shares and distributions tied to redeemed shares. It is the exact line that tells you whether BlackRock’s IBIT and ETHA are absorbing new institutional capital or returning old capital.
The Aug. 6 filings for the iShares Bitcoin Trust ETF (IBIT) and the iShares Ethereum Trust ETF (ETHA) do not name the parties on the other side. But the receipts are plain. 106,148 BTC. 770,839 ETH. Those numbers sit in a row labeled “assets sold for share redemptions.”
In my two decades reading smart contracts and SEC forms, I have learned one rule. When money moves quietly at the trust level, the market usually feels it months later. This is not the kind of rule that makes a good tweet. It is the kind of rule that pays when the crowd is staring at a candlestick instead of a footnote.
Let’s set the baseline. The capital-share line on a U.S. ETF filing is not a holdings update. It is not a wallet transaction. It records contributions tied to shares issued minus distributions tied to shares redeemed. It is deliberately separated from price-driven changes in the trusts’ net assets and from investors’ individual profits or losses.
That separation is critical.
Bitcoin price movements can reduce net assets even if no holder sells a single token. Conversely, capital-share activity can appear massive without any cryptocurrency touching a public order book. In-kind redemptions, authorized participants, market-maker inventory shifts, and cross-custody transfers all generate filing data that looks like selling to the untrained eye.
The numbers are stark.
IBIT recorded $4.3 billion of contributions for shares issued and $7.2 billion of distributions for shares redeemed during the three months ended June 30. The difference produced a $2.9 billion net decrease.
ETHA recorded $943.3 million in contributions and $1.5 billion in distributions. That produced a $583.4 million decrease.
Combined: $3.5 billion.
The prior-year comparison is the real headline. The 2025 IBIT filing and the 2025 ETHA filing show a combined $13.9 billion increase in the same quarter. That is a $17.4 billion year-over-year swing.
Let that sink in. In Q2 2025, the market poured net capital into these two products at a rate of roughly $150 million per day. In Q2 2026, the market reversed the direction and pulled out $3.5 billion over 63 days. If you are looking for a registration of institutional demand, that is a dramatic pivot.
But it is not automatically a collapse. The same trusts that saw $13.9 billion inflow in the prior-year quarter were also sitting on enormous unrealized gains. When the market consolidates and price momentum fades, profitable holders are the first to redeem. That behavior is not uniformly bearish. It can be market structure noise, tax optimization, or a custody change.
The filings contain one more detail that compounds the bearish-looking picture. IBIT’s operations reduced net assets by over $7 billion during the second quarter. ETHA’s reduced them by $1.5 billion. Those totals include net realized losses and unrealized depreciation at the trust level. So the trusts shrank by $8.5 billion even though the capital-share net decrease was only $3.5 billion. The difference was price.
This is where the forensic work starts.
The activity tables placed 106,148 BTC and 770,839 ETH in rows labeled as assets sold for share redemptions. Every headline writer will use those numbers to claim BlackRock is dumping cryptocurrency. But the footnotes contain a qualifier that changes the analysis.
The footnotes say those rows include in-kind distributions valued at $3.85 billion of Bitcoin and $904 million of Ethereum, without disclosing the unit-level split.
What does that mean in practice?
An ETF can redeem shares for cash or in-kind. In a cash redemption, the trust sells the underlying asset and sends the proceeds. In an in-kind redemption, the trust hands the actual Bitcoin or ether to the authorized participant. The AP then owns those coins. The trust no longer does. Whether the AP sells into the open market, places them on OTC desks, moves them to a new custody wallet, or holds them as principal inventory is not disclosed in the filing.
That is the key nuance. The full token quantities cannot be treated as wholly open-market sales. They are not exchange fill data. They are custody events.
I have seen this pattern in multiple market cycles. It is the same reason I publish direct links to Etherscan when I break a wallet-cluster report. Raw numbers on a form are the beginning of an investigation, not the conclusion.
A key layer is trust-level NAV destruction. IBIT operations reduced net assets by more than $7 billion. That is a massive line item. It includes net realized losses and unrealized depreciation. If Bitcoin spent most of Q2 trading below the average cost basis of recently issued shares, every redemption crystallized a loss at the trust level. The $7 billion is a mix of actual losses paid out and the devaluation of remaining coins. It is not pure selling.
Another layer is the share-count reversal. IBIT’s $4.3 billion in contributions against $7.2 billion in distributions means the trust lost share count. But there is a temporal mismatch. Contributions happened at one BTC price. Redemptions happened at another. The capital-share line is denominated in dollars, not in coin quantity. Without the unit-level split for in-kind distributions, we cannot reconstruct the exact BTC price at which each redemption occurred.
The third layer is distribution-flow timing. The $3.85 billion of in-kind Bitcoin distributions and $904 million of Ethereum in-kind distributions are components of the row labeled assets sold for share redemptions. The unit-level split is not disclosed. That missing detail is material. A cash redemption requires the trust to sell Bitcoin on the open market or via block trade. An in-kind redemption does not.
The fourth layer is the authorized participant’s role. In-kind redemptions transfer the inventory burden from the ETF to the AP. The AP can hold, sell, or hedge. Large APs often execute pre-arranged block trades with OTC counterparties, which bypass visible exchange order books. So even if the 106,148 BTC were ultimately sold, the sale is not a time-stamped market dump that appears in Coinbase order flow. The price impact path is opaque.
This is the kind of situation that demands a custom script. During my 2020 audit of DeFi liquidity pools, I built a system to scrape governance votes and cross-reference them with Uniswap LP balances. That work taught me that on-chain labels are often a starting point. The same applies here. The label “assets sold for share redemptions” is not the same as a market sale.
Let me be more precise with the line-by-line numbers.
For IBIT: - Contributions for shares issued: $4.3 billion - Distributions for shares redeemed: $7.2 billion - Net capital-share decrease: $2.9 billion - Assets recorded as sold for share redemptions: 106,148 BTC - In-kind distribution value included: $3.85 billion
For ETHA: - Contributions for shares issued: $943.3 million - Distributions for shares redeemed: $1.5 billion - Net capital-share decrease: $583.4 million - Assets recorded as sold for share redemptions: 770,839 ETH - In-kind distribution value included: $904 million
Combined net decrease: $3.5 billion Prior-year combined increase: $13.9 billion Year-over-year swing: $17.4 billion
The proportions matter. The in-kind distribution value for IBIT is $3.85 billion against a total distribution line of $7.2 billion. That means at least part of the distribution flow was conducted without selling Bitcoin into an open market. For ETHA, the in-kind value is $904 million against $1.5 billion. Again, a significant portion of the distribution flow bypassed the raw exchange order book.
None of this tells you the APs decided to hold. They could have received the in-kind coins and sold them into the market an hour later through a dark pool. But you cannot assume that from the filing. That is the difference between forensic reading and headline reading.
There is a limited counterweight in the current data.
As of Aug. 6, Farside Investors’ latest completed Bitcoin ETF row showed a $196.8 million IBIT inflow on Aug. 5. Its Ethereum ETF table showed $50.3 million for ETHA. Across Aug. 3-5, IBIT captured $478.5 million in inflows, and ETHA drew $83.8 million.
Total across those three sessions: $562.3 million.
Against the Q2 net decrease of $3.5 billion, that is meaningful at the margins but small in scale. As a nominal scale marker only, $562.3 million equals 15.9% of $3.5 billion. In other words, August has recovered less than one-sixth of the Q2 net redemption.
If August sustains the same $187.4 million combined daily average, it would take about 19 trading sessions for BlackRock funds to accumulate a similar amount. That is approximately four weeks. That shows why persistence over weeks is the more meaningful test. A single strong day after a quarter of redemptions is not a trend. A month of consistent daily inflows is.
The obvious takeaway is bearish. A $3.5 billion net decrease. 106,148 BTC in the “assets sold” row. 770,839 ETH. Add those headlines and retail reads “BlackRock is dumping crypto.”
I think that framing is flawed.
Let’s start with accounting. The capital-share line is a plumbing mechanism for ETF creation and redemption. It is not an investment thesis. A net decrease of $3.5 billion in a quarter can be produced by a small number of large APs rebalancing their share inventory after a massive prior-year run. When Bitcoin rallied during the ETF approval cycle, many institutional investors held significant unrealized gains inside IBIT. Q2 2026 gave them a chance to realize gains with tax efficiency, especially if they could redeem in-kind and avoid book capital gains. That is a profit-taking event, not an asset-class exit.
The prior-year quarter was the outlier. $13.9 billion of net issuance in Q2 2025 was extraordinary. That quarter captured a global wave of approval-driven demand. If you benchmark every subsequent quarter against that frenzy, every normal quarter looks weak. But net redemptions of $3.5 billion in a consolidating, choppy market is not capitulation. It is normalization.
The in-kind redemptions are the third reason to push back on the dump narrative. The row includes $3.85 billion of Bitcoin in-kind distributions. When a trust distributes Bitcoin in-kind, there is no immediate open-market sale. The AP can hold the Bitcoin. It can sell it into a private block trade. It can shift it to an OTC desk. None of those channels are visible in the filing. So citing 106,148 BTC as “sales” overstates the realized open-market supply.
The fourth reason is the counterparty mystery. The filing does not identify who initiated the redemptions. It could be a large asset manager rebalancing its crypto allocation. It could be a hedge fund unwinding a basis trade. It could be a pension plan switching to direct custody. Each of those scenarios has a different price impact and a different future behavior.
During the FTX collapse in 2022, I analyzed the public Solana ledger within the first 48 hours. I found $1.2 billion in hidden transfers to Alameda Research accounts. The key lesson was that wallets are not people and labels are not motives. A transfer to an address labeled Alameda looked like insider activity, but some of those transfers were collateral movements, not sales. In the same way, a row labeled “assets sold for share redemptions” is not equivalent to market dumping.
There is also the cost-basis issue. The $3.5 billion net decrease is in dollars, but the trusts held Bitcoin and Ethereum. If Bitcoin traded lower during Q2, a redemption of 10,000 BTC is recorded at a lower dollar value than a redemption of 10,000 BTC in Q2 2025. The dollar swing overstates the coin-level change. We do not have the exact per-unit split for in-kind distributions, so we cannot fully separate the price effect from the quantity effect. That is another reason to be cautious before declaring an institutional exodus.
None of this means the redemptions are harmless. If the redeemers were market-neutral arbitrage desks unwinding ETF premium trades, they could be selling the underlying coins to neutralize their delta. That creates real pressure on the spot market. The filing cannot distinguish that from a long-term holder moving coins to cold storage. The uncertainty is real. But uncertainty is not the same as bearish.
In 2017, during the ICO audit sprint, I looked at Golem’s smart contract allocation and saw a mismatch between the token generation event and the vesting schedule. The whitepaper said one thing. The code said another. The market celebrated while I was short. The lesson: the headline is not the document. This filing is the same. The headline is “BlackRock redemptions.” The document is a capital-share line with a footnote. The winner is the reader who reads the footnote.
The market is in a sideways phase. Choppy price action means flows matter more than price. Every dollar of net creation is a signal of demand that can later turn into buying pressure. Every dollar of net redemption is a signal that has to be absorbed by existing liquidity. This filing tells us the second signal dominated Q2. August tells us whether that signal is extending.
In a sideways market, a redemption wave is more dangerous than in a bull market. In a strong uptrend, every redemption is met by new creation. In a chop, the APs are less willing to recycle those shares. The Q2 redemptions therefore had a disproportionate effect on sentiment. They were cited as the reason Bitcoin could not break out. But that causal read is lazy. The capital-share line is a lagging metric. It records what happened. It does not predict what happens next.
Institutional flows are rarely linear. A redemption wave often exhausts itself after large holders reposition. The August inflows are early evidence that the wave is settling. But early evidence is not enough. We need to watch the composition of those inflows. Are they coming from new APs? Are they concentrated in a single day? If a single market maker is creating and redeeming to capture bid-ask spread, the daily flow numbers can be inflated without reflecting durable demand.
I have seen this in the 2020 yield farming cycle. Television reporters looked at TVL numbers and called it demand. Anyone who read the governance votes and LP minting timestamps knew that many of those metrics were the same capital looping through protocols. The same is possible in the ETF flow tables. The 19-day persistence test filters out churn.
This filing also exposes a structural weakness in how the market reads ETF data. The same number can be cited to support bullish and bearish conclusions. $4.3 billion in contributions is real buying. $7.2 billion in distributions is real selling. But without reconciling the counterparties and the timing, the same data can be used by either side. That is why I keep emphasizing the footnotes. The footnotes are where the honest part of the analysis lives.
After years of doing this, I have learned to respect the parts of a filing that are absent.
We know the aggregate value: $4.3 billion in, $7.2 billion out for IBIT; $943.3 million in, $1.5 billion out for ETHA.
We know the token quantities: 106,148 BTC and 770,839 ETH.
We know the in-kind distribution values: $3.85 billion BTC, $904 million ETH.
We do not know the unit-level split between cash and in-kind. If cash redemptions dominate, the trust sold far more Bitcoin into the open market than the in-kind number suggests. If in-kind redemptions dominate, the actual market pressure is low. Without the unit-level split, any conclusion about “selling” is incomplete.
We also do not know whether the redeemed shares were original seed shares from the trust’s creation or newer shares bought at higher prices. Original seed shares have a low cost basis. Redeemers of those shares are more likely to be institutional desks with deep liquidity connections. Newer shares are more likely to belong to momentum funds. The cost basis influences what they do next.
Finally, we do not know the destination wallets for the tokens distributed in-kind. That information would allow me to track whether the coins moved to exchanges, to OTC firms, or to custody addresses. I built a similar wallet-cluster tracker during the NFT floor-price manipulation investigation in 2021. It took hours to trace $4 million in wash trading. The same approach could, in theory, trace a portion of this redemption flow. The filing, however, does not hand us the addresses.
That is the next investigative step. If the market really wants clarity on BlackRock’s flow, the largest APs and their counterparties will eventually leave fingerprints on-chain. It may not happen immediately. But the ledger is the truth.
Here is how I am framing this for the coming month.
The Q2 filing tells us the past. August tells us the present. The $562.3 million inflow recorded across Aug. 3-5 is a start. It counters roughly one-sixth of the Q2 net redemption. To fully offset $3.5 billion, the current $187.4 million combined daily average must continue for about 19 trading sessions.
That is the threshold I am watching.
If August can deliver 19 sessions of sustained creation, then Q2 redemptions were a rebalancing event and not an institutional exit. If inflows fade after two weeks, then the capital-share reversal has lasting momentum.
I do not need to know the names of the mystery investors to understand what they did. 106,148 BTC and 770,839 ETH moved out of the trusts. That is a fact. The open question is whether those coins returned to the market or to a different shape of custody.
Code doesn’t lie. But the stories we build around code can. Read the footnotes. Follow the flows. And do not mistake a custody event for a market sale.
The chain does not care about your thesis. It only cares about your receipts. We have the receipts. Now we wait for the next block.
