The Drained Reservoir

A peg is a promise about a buffer

STRC — Strategy's Variable Rate Series A Perpetual Stretch Preferred — is built to trade at its $100 stated amount. The design has three levers: an at-the-market program that issues new shares when the price is above par, a dividend rate management adjusts monthly (11.50% in June, 12.00% since July), and a US-dollar reserve that stands behind the dividend. By June 30 the class had grown to 104.9 million shares — $10.5 billion of notional — making it the largest of Strategy's five listed preferreds and the one carrying the heaviest coupon bill: $282.9 million of the $400.2 million in preferred dividends the company paid in the second quarter alone.

The reserve is the load-bearing lever, and it launched on December 1, 2025 sounding firm: an initial $1.44 billion, with a stated intention to hold at least twelve months of dividend and interest coverage and “the goal of ultimately covering 24 months or more.” But the same announcement kept the reserve's terms and amount subject to Strategy's “sole and absolute discretion” — and by the time the balance resurfaced in May, the filings had quietly downgraded the description to “a management-designated portion of Strategy's liquidity.” The language softened as the balance shrank.

For a while the design worked exactly as drawn. From late February to mid-May 2026, STRC closed between $99.29 and $100.06 — pinned to par for the better part of three months, with early May spent within a few cents of $100.00.

The drawdown nobody watched

The weekly 8-Ks disclosed the reserve balance twice around the turn of the year — $2.19 billion as of December 21, $2.25 billion as of January 4 — and then went quiet. For twenty weeks, no balance appeared in any filing. When the number resurfaced in a May 26 capital-structure update, it was $871 million: down 61%. Against the roughly $1.76 billion Strategy was by then paying annually in preferred dividends and interest, the buffer covered about six months — for a security whose entire premise is that the dividend is safe.

The tape had already wobbled — STRC slipped from $99.19 to $98.61 in mid-May before steadying near $99.50 — but the sustained break dates from the disclosure. STRC closed at $98.57 on May 28, two sessions after the $871 million print, and never closed above $99 again. The slide ran through June with mechanical logic: a preferred stock's par peg is only as strong as the buffer behind it, and the buffer was going the wrong way.

Gravity did the rest

June turned the drift into a cascade. On June 18 STRC closed at a then-record-low $88.59 on 10.8 million shares — the heaviest session of the entire slide — and Strategy's STRC at-the-market issuance stopped: every subsequent weekly filing shows zero shares sold, with $17.5 billion of authorized capacity sitting unused. The ATM lever only works above par; below it, the same program that once absorbed excess demand becomes a standing threat of supply. By June 26, amid a broad drawdown across the Strategy complex — MSTR closed that day at $82.31, STRD had closed at $50.00 the session before, and the company would book an $8.32 billion quarterly loss on its bitcoin holdings — STRC closed at $74.57, a quarter below its stated amount.

The refill, governed this time

The June 29 answer — the Digital Credit Capital Framework — reads as a point-by-point repudiation of the drawdown. The reserve got a Board-approved policy: it “may be used only to support the payment of preferred stock dividends and interest expense,” any other use requires Board authorization, and it carries a floor of twelve months of coverage. The STRC dividend went to 12.00%. The Board authorized $1 billion of preferred repurchases and $1 billion of common repurchases, and a bitcoin monetization program permitting up to $1.25 billion of BTC sales — under 2.5% of holdings — primarily to fund the reserve itself, with dividend service and repurchases as subordinate uses.

The market graded the announcement, not the execution. STRC jumped from $74.57 to $88.58 in the seven sessions after the framework — +$14.01 of the eventual +$20.44 recovery, 69% of the whole move, on zero buybacks. Then came the cleanest natural experiment in the story: between July 6 and July 17 the reserve climbed from $2.55 billion to $3.225 billion while STRC fell 3.7%. The refill, on its own, moved coverage from 17 toward 22 months and moved the price down. Whatever was repricing STRC, it was not the reserve balance per se.

The reserve kept climbing — $3.75 billion by July 26, $4.0 billion by August 2, roughly 27 months of coverage — funded by MSTR common ATM sales (over $1.5 billion across July) and selective bitcoin sales. The rebuild matters enormously for the credit; it simply is not the thing the tape was paying for week to week.

What the buybacks actually did

The first repurchases came July 20–26: 288,930 shares for $25.0 million, about 5.5% of that week's volume. The next week tripled it: 912,143 shares for $81.2 million, 13.9% of weekly volume — funded, notably, by $52.3 million of bitcoin-sale proceeds and $28.9 million of MSTR ATM proceeds — the reserve's policy limits it to dividends and interest, and any other use would require specific Board authorization, which has not been given. Cumulatively that is 1.15% of the class and 10.6% of the authorization.

Did it move the price? We regressed STRC's daily returns on STRD and STRF — same issuer, same credit, and zero repurchases, though both were eligible under the same authorization; management simply bought only STRC — over the five weeks before the buyback began (R² = 0.90) and measured the abnormal return during the buyback window. The headline answer says yes: a cumulative abnormal return of +12.6% (t = 2.5). But the result is fragile. Force the estimation alpha to zero and it falls to +6.6% (t = 1.3); also drop July 31 and it collapses to +1.8% (t = 0.4). Nearly the entire measured effect is one session — July 31, when the dividend was held at 12% and STRC closed flat while STRD fell 3.8% and STRF 3.4%. That single print is the best evidence the bid does something: it absorbed a complex-wide selloff. It is also one observation.

The final leg makes the point from the other side. From August 3–7, STRC rose 6.2% — while STRD, with no buyback support at all, rose 14.0%. The sibling without the bid outran the sibling with it. That is beta to a recovering complex, not a buyback premium.

The ranking

1. The reserve drawdown started it. A 61% depletion of the dividend buffer, disclosed after a twenty-week gap, is the proximate trigger for the peg break.
2. The June 29 framework reversed it — roughly two-thirds of the recovery, priced on announcement. Governance was the product: the same dollars, ring-fenced, bought a different multiple.
3. The ATM shutoff quietly removed a standing seller. Zero issuance since mid-June against $17.5 billion of capacity is a larger flow change than any buyback week.
4. Complex beta — MSTR +21% and STRD +32% off their late-June lows — carried STRC with it.
5. The buybacks rank last: small against the class, real as a signal (selling bitcoin to defend the preferred is costly, hence credible), fragile as a measured price effect.

Two things bound the trade from here. STRC is redeemable at the company's option at $101, so “near par” is the ceiling by design — this is credit convergence, not upside. And the week of August 3–7, the strongest of the recovery (+6.2%), sits in a disclosure window that had not yet been filed as of this writing; the next weekly 8-K will show whether the bid was even present for it.

The reserve, as disclosed

Addendum · August 10, 2026 — the bid was present

The disclosure window this note flagged as open has now been filed. The August 10 8-K covers August 3–9 and answers the closing question directly: the bid was present — and it was the largest yet. Strategy repurchased 1,152,020 STRC shares for $108.6 million, the third and biggest buyback week, taking the cumulative program to 2,353,093 shares and $214.8 million — about 2.2% of the class and 21.5% of the $1 billion authorization, with $785.2 million remaining.

The funding mix repeated the pattern, at greater cost. The entire $108.6 million was funded by bitcoin sales: 1,690 BTC at an average $64,262 — roughly 15% below the company's $75,385 average purchase price — leaving holdings at 840,447 BTC. The reserve, meanwhile, climbed to $4.65 billion as of August 9 — roughly 32 months of coverage on the company's $1.76 billion annual figure, by our arithmetic — funded by $650.0 million of the week's $653.1 million in MSTR common ATM proceeds (6,585,682 shares sold). The STRC ATM stayed at zero for an eighth consecutive week. The division of labor is now explicit in the filing's own footnotes: common-stock dilution refills the reserve; bitcoin sales fund the buyback.

The new week strengthens the ranking rather than revising it. The strongest week of the recovery did coincide with the largest buyback — but STRD, with no bid at all, rose 14.0% in the same sessions against STRC's 6.2%, so the beta reading stands. The signal, however, keeps getting costlier and therefore more credible: selling bitcoin below its average purchase price to retire preferred stock at a discount to its stated amount is exactly the sort of expensive commitment the June 29 framework promised. And day one graded the execution the way execution weeks have been graded all along: STRC closed at $94.35 on August 10, off 0.7% — the filings that reprice this security are announcements, not receipts.