The Aug. 4 Form 10-Q is not a press release. It does not celebrate. It reports. And the report contradicts the narrative that circulated through crypto media in the first week of July. Grayscale’s XRP Trust — ticker GXRP — recorded 510,000 shares issued and 30,000 redeemed during the second quarter of 2026. Net creations: 480,000. The outlets called it a comeback. The ledger calls it a 12.2% clawback of a prior contraction. One share recovered for every eight lost in the preceding three months. At June 30, 87.8% of that contraction remained unrecovered. This is not a recovery. It is the statistical echo of a balance sheet that is still bleeding.
I have spent eighteen years auditing this industry’s books. The FTX collapse taught me one durable lesson: the space between a funded narrative and a funded account is where the truth lives. GXRP’s six-month ledger is that space, made visible, in numbers that flatter no one.
Context: A Trust Under Inspection
GXRP is a single-asset exchange-traded product holding XRP, one of several vehicles that emerged after the asset’s regulatory status stabilized in the mid-2020s. The product’s trajectory through 2025 was genuinely strong. Institutional appetite for XRP exposure, driven in part by Ripple’s diversified offerings and ongoing XRPL development, pushed the fund to a high-water mark of 122.230 million XRP held as of Dec. 31, 2025. The first half of 2026 has been a different regime.
The broader ETF complex tells the same story in a different key. Bitcoin and Ethereum ETFs lost roughly $2.5 billion through June 18, while HYPE and XRP funds drew less than $75 million combined. Analysts described the behavior as de-risking rather than rotation. The XRP-specific flows flipped from inflows to outflows in March, after a $1.2 billion run. By the time GXRP filed its June 30 report, the question was not whether the fund had suffered. The question was how badly — and whether one quarterly positive print could reverse the damage.
The answer, in raw shares, is: not yet. The six-month statement of changes in shares outstanding shows 1.87 million shares issued and 5.33 million redeemed, a net contraction of 3.46 million units. Subtract the second quarter’s activity from those totals, and the first quarter’s derived numbers emerge: 1.36 million shares issued, 5.30 million redeemed, and a net contraction of 3.94 million shares. The Q2 clawback of 480,000 shares recovers only 12.2% of Q1’s derived contraction. History is the only reliable audit trail, and the trail here is unambiguous.
The derivation itself is worth noting. The 10-Q reports period aggregates. To isolate Q1, I subtracted Q2’s reported issuance and redemption figures from the six-month totals. That is standard SEC arithmetic. It is also, in this case, the only way to see Q1’s true depth, because the filing does not give it to you directly. My work on L2 fraud proof benchmarks taught me a rule that applies far beyond scaling protocols: when a system reports only aggregates, calculate the residual. The residual is where the signal lives.
The trust’s structure makes the filing unusually legible. A single-asset grantor trust holds XRP in custody, issues shares against in-kind deposits, and redeems shares by distributing XRP or its cash proceeds. There are no derivatives, no lending programs, no yield wrappers. Every dollar in the fund is a token, and every token is disclosed. That transparency is the fund’s only grace. It is also why the numbers are so damning: there is no complexity to hide behind.
Core: The Forensic Breakdown
The Share Ledger Does Not Lie
Start with the share ledger because it is the most objective statement in the filing. Q2: 510,000 shares issued, 30,000 redeemed, net creations of 480,000. That is the only positive quarterly net-creation figure in the 2026 record. Praise it with faint damns.
Now place it next to Q1. The derived Q1 figures describe a run on the fund: 1.36 million shares issued against 5.30 million redeemed, a net contraction of 3.94 million shares in thirteen weeks. The Q2 recovery of 480,000 shares replaces approximately 12.2% of what Q1 removed. The unrecovered balance at June 30: 3.46 million shares, or 87.8% of the initial wound.
There is a proportional framing the press missed. The 480,000 shares created in Q2 represent 16.9% of the fund’s 2,840,100 shares outstanding at quarter-end — a nontrivial increase in the unit base. But measured against the 3.94 million shares that exited in Q1, the same number collapses to 12.2%. Both figures are mathematically true. Only one of them tells you the fund is still 87.8% below its March starting line.
Words matter in this industry because the industry runs on narrative. The term “clawback” was attached to the Q2 figure in the related reporting. A clawback, in the legal sense, is a recovery of previously distributed value. What actually happened in Q2 is issuance: new shares sold to new or existing holders. No value was clawed back. Value was supplied. The distinction is not pedantic. It changes whether the $12.743 million is a return of departed capital — it is not — or fresh capital entering a wounded vehicle. It is the latter.
Capital Added, Then Eaten
Q2 share activity added $12.743 million of net capital. Issuances brought in $13.442 million; redemptions paid out $699,000. In any ordinary fund quarter, $12.7 million of net capital is a healthy print. But the fund does not operate in a vacuum. The $16.846 million loss from operations overwhelmed the capital contribution. Net assets fell from $61.516 million at the end of March to $57.413 million at the end of June. The new money arrived, and the market took it away, plus $4.1 million in change.
Gross flows tell a sharper version of the same story. The asymmetry between quarters is an order-of-magnitude event: Q1’s derived redemption of 5.30 million shares, at Q1 unit values near $21, implies an exit measured in the nine figures, against a Q2 gross issuance of $13.4 million. The quarterly flows are not converging. They are oscillating around different baselines.
This is the quarter’s central fact: capital creation added, investment performance subtracted, and subtraction won.
The $16.789 million realized and unrealized investment loss decomposes into three components. A $16.327 million drop in unrealized appreciation dominates. $433,000 in realized losses came from XRP sold to satisfy redemptions. $29,000 in realized losses came from XRP sold to pay expenses. A separate $57,000 net investment loss — the fund’s fee and expense drag, roughly 38 basis points annualized against average assets — rounds the operating decline to $16.846 million.
Dissect the unrealized number. A $16.327 million decline in unrealized appreciation means the XRP the trust holds fell in mark-to-market value during the quarter, even as the trust’s XRP balance grew 20.2%, from 45.774 million to 55.036 million tokens. The fund accumulated more XRP, and the XRP it held lost dollar value. That is the double bind of a single-asset trust in a falling market: you can add units and still lose net assets.
The NAV math exposes the underlying price action without requiring a single price feed. At March 31, the fund’s $61.516 million in net assets against 45.774 million XRP implies a per-token valuation of approximately $1.34. At June 30, $57.413 million against 55.036 million XRP implies $1.04. Derived from the trust’s own books, XRP lost roughly 22.3% of its dollar value in three months. The trust is, in effect, a clean price oracle for the asset — and the oracle says the asset bled through the entire quarter. Data does not negotiate; it only confirms.
Redemptions: The Double-Exit Structure
The $433,000 realized loss on XRP sold for redemptions deserves separate attention. This is not a theoretical mark-to-market adjustment. This is the fund selling tokens at prices below its average acquisition cost to satisfy shareholder exits. Every redemption in Q2 forced the monetization of corpus into a market trading below the trust’s embedded basis.
The same mechanism operates at smaller scale with expenses. The $29,000 realized loss on XRP sold to pay operational costs — audit, custody, legal, sponsor fees — shows that even the mundane plumbing of fund administration is running at a loss in token terms.
Here is the structural point the bulls ignored. A redemption in a single-asset trust is a double exit. The departing shareholder takes cash, reducing net assets by the redemption value. The fund then realizes a loss when it sells XRP into a depressed market, reducing net assets a second time. The staying shareholders absorb both blows. In a declining market, every redemption taxes the remaining investors twice. My 2024 stablecoin depeg work surfaced the same dynamic: when the marginal buyer disappears, the cost of exit is borne by those who remain. Liquidity depth is the tell. In GXRP’s case, the tell is the $433,000 realized loss — small in absolute terms, devastating as a signal of what redemptions cost a trust in a down market.
The First-Half Aggregate: $165.951 Million in Context
The first-half statement shows a total decline in net assets of $165.951 million. Capital-share transactions accounted for $114.203 million of that decline; operations for $51.748 million. Both engines of destruction ran in the same direction for six consecutive months.
Read that allocation carefully. Sixty-nine percent of the fund’s first-half net-asset decline came from shareholder departures, not from market depreciation. Unrealized losses can recover when XRP rallies. Redeemed shares are gone forever. Redemptions are a structural vote of no confidence, expressed in action rather than commentary.
The ratio deserves emphasis. GXRP paid out $114.203 million in net redemptions during the first half — nearly twice the $57.413 million in net assets the fund still holds. In two quarters, shareholders withdrew an amount roughly equal to two full copies of everything left in the box. That is not a correction. That is a liquidation curve in its early stages. The trust survived because new creations, however modest, offset part of the drain. The asymmetry between Q1’s nine-figure exit and Q2’s $12.743 million entry is the entire story of the product’s first half.
July’s Static Share Count
GXRP reported 2,840,100 shares outstanding on both June 30 and July 30. The reconciliation: any creations and redemptions during July offset each other exactly by the end of the month. Two readings are possible. The optimistic reading: the exit wave has exhausted itself, and the fund has stabilized. The pessimistic reading: the entry wave has not begun, and the fund is parked at a plateau.
My forecasting work on algorithmic stablecoin reserves showed me that stability is not the same as safety. A reserve ratio that stops falling is not adequate; it is merely no longer deteriorating. The same logic applies here. A share count that stops falling in July is not evidence of renewed demand. It is evidence that the selling pause matched the buying pause. The broader market context supports the cautious read: the XRP complex cleaned out its leverage in June, and spot volume plus ETF creations carried the subsequent move. But in GXRP’s case, the carry was zero. Net. Zero. For a month.
What the Fund Holds Now
The XRP balance grew 20.2% in Q2, from 45.774 million to 55.036 million tokens. Read against the December baseline, however, the picture darkens. Holdings finished June 55.0% below the 122.230 million XRP held on Dec. 31, 2025. The trust holds less than half of its year-start corpus.
The mechanics of that growth matter. The 9.262 million XRP increase in the trust’s holdings corresponds almost exactly to the 480,000 net shares created at an implied unit NAV of roughly $20.20 — a $9.7 million in-kind contribution. The balance grew because creations brought tokens in kind, not because the sponsor bought the dip. There is no discretionary buying; there is only a pass-through reflecting the net of investor entries and exits. Proof is cheaper than trust, yet still ignored: the fund’s enlarged XRP balance is a mechanical function of its net creations, and its net creations are a rounding error against its net redemptions.
A Note on Fee Drag
The $57,000 net investment loss is the quietest number in the filing. Against average net assets of approximately $60 million, it implies an annual expense ratio in the range of 38 basis points. That is modest by ETF standards. But in a trust that is contracting, the fee is paid from a shrinking base, and the $29,000 realized loss on XRP sold to pay expenses converts that fee from a cash charge into a token disposal. Every quarter of contraction makes the trust’s operating overhead slightly more corrosive, because a portion of it is paid by selling assets at a loss. Silence in the code is a bug waiting to happen; silence in the fee schedule is a drain waiting to compound.
What the Filing Withholds
For all the transparency, the 10-Q leaves critical data in the dark. It does not disclose the percentage of creations and redemptions executed in-kind versus in cash. It does not name the authorized participants. It does not break the XRP sales into discrete transactions, so the exact average sale price for the units sold to satisfy redemptions is unknowable from the public record. And it does not reveal the fund’s cost basis per XRP, which means the size of the embedded capital loss on remaining holdings is a matter of inference.
That opacity matters for risk managers. In my FTX work, the fatal gap was between what the exchange claimed to hold and what its reserve proofs actually showed. Here, the gap is smaller but real: investors cannot fully reconstruct the trust’s tax position, its liquidation cost curve, or the identity of the counterparties routing the redemptions. The static July share count tells us that offsetting activity happened; it cannot tell us who was on either side. In a product that has lost 69% of its first-half decline to shareholder exits, knowing who is leaving is as important as knowing how many.
Contrarian: What the Bulls Got Right
I am not going to pretend the second quarter was an unbroken sequence of red flags. The bulls can point to five defensible facts.
The strongest is the issuance ledger. 510,000 shares were created in Q2. Someone created. In a market where bitcoin and Ethereum ETFs shed $2.5 billion through mid-June, a single-asset trust with XRP exposure attracted $13.442 million of fresh issuance. That is the strongest demand signal the XRP-linked product complex showed in the first half. It is not nothing.
The token balance supports them too. XRP holdings grew 20.2%, improving the token-per-share ratio for remaining holders. If the price recovers, the quarter’s creations will have positioned the corpus for more upside per unit than the March configuration offered.
The static share count from June 30 to July 30 suggests the redemption wave has at least paused. A fund that stops shrinking is a fund that has time. Time is the only asset that matters in a sideways market.
The realized losses were small. $433,000 on redemptions and $29,000 on expenses are rounding errors on a $57 million fund. The damage to net assets in Q2 came predominantly from unrealized depreciation — a reversal that would retrace if XRP mean-reverts. The bull case is not absurd; it is merely incomplete.
The institutional adoption thesis deserves credit it is not receiving. The fund’s survival through a six-month drawdown is itself evidence of a durable shareholder base. In the ETF graveyard of the past decade, dozens of single-asset trusts liquidated under far milder redemptions. GXRP kept its sponsor, custody, and listing. That is not nothing in a market that de-risked everything else.
The blind spot is the framing. A positive quarterly print on the share ledger is not a recovered fund. The headline — “Investors poured millions back into Grayscale’s XRP fund, but a $16.8M market hit instantly wiped out the comeback” — got the arithmetic right and the conclusion wrong. The $16.8 million hit did not wipe out a comeback; the $12.743 million injection was never a comeback to begin with. It was a partial offset. Net assets fell at the end of the quarter the fund supposedly “won.” Consensus is not a feature; it is the foundation. The consensus here was built on a share count alone, ignoring the dollar statement in the same filing.
Takeaway: The Next 10-Q Is the Test
The second half of 2026 will settle the question. If GXRP posts another quarter of net creations — meaningfully more than 480,000 shares — and XRP holds above its June implied value of roughly $1.04, the recovery thesis gains real weight. If the third quarter repeats the first-half pattern — small issuances, persistent redemptions, negative operating results — the Q2 print will be remembered as exactly what the arithmetic says: a 12.2% clawback of a 3.94 million-share wound. I would add a third scenario the bulls have not priced: a quarter of small net creations with declining XRP prices, producing another round of realized losses on redemption sales. That combination would leave the trust near 40 million XRP by year-end, with net assets below $40 million if the token holds its June level. The pattern is the projection.
The Aug. 4 filing was public the entire time. The comeback was written before the ledger was read. I have watched this sequence repeat — in FTX’s reserve proofs, in stablecoin collateral ratios, in every product that marketed its way through an unwinding while its balance sheet told a different story. The ledger does not lie, only the operators do. In this case, the operators did nothing improper. The filing is accurate. It was the interpretation that failed. Read the 10-Q. Do the subtraction. The numbers are right there, and they do not say what the headlines said.