Disclosures & methodology

Not affiliated with Strategy or Michael Saylor

Saylor's Accountant is an independent project by Dave Lawler at Velocity Point. It is not affiliated with Strategy Inc, MicroStrategy Incorporated, Michael Saylor, or any officer, employee, or representative of Strategy. The name is rhetorical — we keep books on Strategy's machine the way an outside accountant would, not the way Strategy itself does. Strategy has not endorsed this site, has no editorial relationship with it, and may not even be aware of it.

The single source of truth for every figure on this site is the SEC filings — primarily Strategy's weekly Form 8‑Ks announcing each Bitcoin purchase and the associated capital raise. Direct links to the underlying filings appear next to the numbers on every page so you can verify any claim against its primary source.

Companion book — Measure in Sats

This site is the live ledger; the book is the long-form theory. Measure in Sats: The Idealized Bitcoin Treasury is a companion book by the same author, working through the math that this site renders against real SEC filings every Monday. The book argues why the mechanism works in idealized form (Forever Cost, accretive dilution, the geometric series, the double-accretion flywheel); the site shows it actually working, week by week, on Strategy's specific capital structure. They share a name because they share a method: price the perpetual obligation in BTC, not in dollars at par.

Each chapter of the book points at the page on this site that extends it empirically. If you arrived here from the book, the appendix maps chapters to pages. If you arrived here cold, the book is where to go for the why behind the numbers.

This is a modeling exercise, not financial advice

The output is a model. It is not:

• A GAAP-compliant financial statement.
• An audit of Strategy's books or an independent verification of its disclosures.
• Investment, tax, legal, or financial advice of any kind.
• A recommendation to buy, sell, or hold MSTR, any preferred series, or any other security.

The model trades exactness for comprehension. It rounds, interpolates between weekly disclosures, projects forward using a single user-adjustable BTC growth rate, and excludes effects that don't change the per-share story (tax treatment for individual dividend recipients, transaction costs, regulatory drag, exchange-rate effects on the EUR-denominated STRE coupon, and so on). For any real-money decision, work from Strategy's audited 10‑K and 10‑Q filings and consult licensed advisors. The forward-looking projections on this site reflect a particular set of assumptions — they are not predictions and should not be relied upon as such.

The model assumes Bitcoin is sold to fund dividends — even though Strategy says otherwise

Michael Saylor has stated repeatedly that Strategy does not sell Bitcoin to fund its dividend payments. That's literally true. Dividends are paid from cash raised through Strategy's at-the-market (ATM) common and preferred stock programs — issuing new shares for dollars — not from Bitcoin sales.

The model on this site nonetheless treats every dividend dollar as the equivalent of selling X BTC at that period's BTC price. That choice is deliberate, and we think it's a more honest framing of the obligation than “we never sell.” Three reasons:

1. The dilution is mathematically equivalent. When Strategy issues new common stock at mNAV > 1 to raise dividend cash, the existing shareholders' fractional claim on the BTC stack shrinks. Whether you describe that as “we sold BTC to pay dividends” or “we diluted shareholders to pay dividends in cash,” the result for an existing shareholder is identical: their fraction of the BTC pile drops by the dividend's BTC equivalent at that moment's price. Modeling sale and modeling dilution-at-mNAV produce the same per-share outcome.
2. The obligation exists regardless of how it's funded. A 10% perpetual preferred owes its coupon every quarter forever. That's a fixed claim on Strategy's economic substance. The substance is the BTC pile. The accounting plumbing — ATM proceeds, retained cash, debt issuance — is interchangeable in the long run; the BTC stack is what either funds the stream or gets diluted by it.
3. Modeling sales makes the geometric series legible. The convergent asymptote that's the whole point of the analysis only becomes a clean equation when every coupon can be expressed as a BTC quantity at that period's price. The BTC-cost page presents two forms: the simple annual-annuity-due D · V · (1+CAGR) / CAGR (a teaching simplification that overstates burn by 12–15% at CAGR=30%), and the cadence-correct ordinary annuity D · V / (k · ((1+CAGR)1/k − 1)) — what Strategy actually faces, with k = payments per year (4 quarterly, 12 monthly). The site's per-tranche and per-pref pages all use the cadence-correct form; the simple form is kept only as the on-ramp explanation. The sale framing is the unit conversion that makes the long-run question (“does the pile survive the obligation?”) answerable in BTC, in either form.

The “we never sell” statement is technically correct and rhetorically useful. The “sold to fund coupons” model is economically correct and analytically useful. We use the second because the question this site is trying to answer — what fraction of the BTC stack survives the perpetual dividend stream — only has a clean answer in that frame.

For the physical cash flow — the actual dollars Strategy raised, spent, paid out in coupons, and held back as a treasury buffer in each fiscal year — see the cash flow reconciliation. That page sources every number from the 10‑K's Consolidated Statement of Cash Flows and shows where the dollars that didn't become BTC actually went (operating costs, interest, taxes, secured-debt principal, and starting in 2025, a deliberate multi-year cash buffer). The atomic-tranche model on this site and the physical cash flow tie out at the balance-sheet level by construction — they're two lenses on the same dollar movements.

“Where does the money come from?” — and why this isn't a Ponzi

The most common criticism of Strategy goes like this: they raise dollars to buy Bitcoin, then they raise more dollars to pay the dividend on the dollars they already raised. If they ever can't raise more, they have to sell Bitcoin. That's a Ponzi. The criticism deserves a serious answer, because it points at a real question — where does the dividend cash actually come from? — but the “Ponzi” conclusion is wrong, and it's wrong for a specific reason worth pinning down.

A Ponzi is a fraud about value. Ponzi schemes pay old investors with new investors' money while telling the old investors their money has earned a return that doesn't exist. The books are cooked. The claimed asset isn't there. When new money stops flowing in, the lie collapses because there was never any underlying. That's the structural feature that makes a Ponzi a Ponzi: cooked books, claimed value that doesn't exist. Strategy's books are not cooked. The Bitcoin is on chain. Anyone with a block explorer can see it. That's not a Ponzi — it's a leveraged play on a public asset.

Where the dividend cash actually comes from is one of two channels, and either way it's accretive:

1. Issuing more common stock (the usual channel). When MSTR trades above its per-share Bitcoin value (mNAV > 1) — which it has for nearly the entire life of this strategy — every share Strategy issues at market is accretive to existing shareholders. They sell new common at $X, where $X is more than the per-share BTC NAV; the cash buys BTC at the spot price; existing shareholders end up with more sats per share than before the issuance. Some of that fresh cash funds the next dividend payment. The math is on the common page.
2. Issuing more preferreds (the alternate channel they avoid optically). Mechanically, Strategy could fund STRK/STRF/STRD dividends with new STRC issuance — the rate-flex preferred is ATMable near par anytime. They don't, because raising preferreds to pay preferreds looks Ponzi-ish even when it isn't. (Mechanically it isn't — see below.) But the option is there if the common-ATM channel ever closed.

Both channels add a future obligation in exchange for cash now. Channel 1 is accretive when mNAV > 1, cleanly and visibly. Channel 2 is accretive when the present value of the future preferred coupon stream — measured in BTC at the BTC-cost asymptote — is less than the BTC the new raise notionally buys. Both are leveraged trades on an asset that compounds. Neither is a Ponzi.

The intuition pump everyone already accepts: imagine a bank offered you a 100-year, 1%-fixed-rate loan. Nobody would say “but how will you ever pay it back?” — you'd take the money instantly, because you can invest in basically anything that returns more than 1% over a century, pay the coupon out of the spread, and pocket the difference. The trade is so obvious that the only reason 100-year 1% loans don't exist is that lenders aren't stupid enough to offer them.

Strategy is taking the same trade, one level up. The “loan” is a perpetual preferred at 8-11% (or a 0% convertible). The “investment” is Bitcoin, which has historically compounded faster than the coupon. As long as that stays true on average over the long run, the geometric series converges: the cumulative BTC ever needed to fund the coupon is finite, and the rest of the BTC stack survives. Whether the dividends get paid out of fresh ATM proceeds (channel 1), new preferred issuance (channel 2), or eventually liquidating BTC (channel 3, never-explicitly-used) doesn't change the underlying economics — it just changes which page of the cash-flow statement the entry shows up on. The substance is the BTC.

What would invalidate the trade: Bitcoin failing to compound faster than the coupon rate over the long run. That's a real risk, and it's the only one that matters. The risk is not “Strategy ran out of buyers for the next ATM” — they have at least three uncorrelated channels and the mechanics work in any of them. The risk is the underlying asset disappointing. Pricing that risk is what investors are doing every day they look at MSTR's quote relative to BTC. The site's job is to make the math under the trade visible, not to advocate for any particular answer to it.

Other modeling caveats

• Forward BTC CAGR is an assumption, not a forecast. The default 30% used across the summary pages is illustrative; the per-tranche detail pages let you slide it. Asymptote calculations are exquisitely sensitive to this number — a difference between 25% and 35% materially changes how much of the stack survives.
• STRC monthly rate resets: the model carries forward the most recently announced rate when no rate-change 8‑K is on file for a given month. Strategy may adjust between filings via its rate card. Months marked with an “interpolated” note in the rate history inherited the prior month's rate.
• Per-leg attribution: each weekly tranche's contribution to each preferred series is parsed from the ATM Program Summary table in the 8‑K. Some weeks fail automated extraction (pdftotext column-wrap is hostile to multi-row description cells) and rely on a manually-curated sidecar override. Where the model lacks per-leg detail, the whole weekly purchase is bundled onto the first-named security, which under-counts the others' contributions for that week.
• Live BTC price is sourced from Coinbase's public spot feed, refreshed every 5 minutes server-side. When the Coinbase API is unreachable the site falls back to the most recent disclosed average price from the latest tranche, which can lag the live market by up to a week.
• EUR-denominated STRE pays its 10% coupon in euros. The model treats it as USD-equivalent at the IPO date's spot rate; subsequent EUR/USD moves aren't reflected. For a euro-investor view of STRE this site under-models the FX layer.

Corrections and contact

Spot a number that doesn't match the underlying 8‑K, or a methodological choice you think we should reconsider? Email Dave at dave.lawler@velocity-point.com. The site is not open source, but we read every note and treat divergence from the primary-source filings as a bug.