Financial Shenanigans

Financial Shenanigans

The reported numbers at Strategy Inc. follow GAAP, but the picture they paint of the business has been engineered. Management blends a cash-burning legacy software franchise with a multi-billion-dollar bitcoin treasury funded by perpetual at-the-market (ATM) equity issuance, then asks investors to judge the result on bespoke "Bitcoin Per Share" KPIs that recognize gains the income statement does not. Reported operating cash flow has been negative for two consecutive years while management distributed roughly $381M in preferred-stock dividends financed entirely by new share sales. Add a controlling founder previously sanctioned by the SEC for accounting fraud (1997-1999) and the District of Columbia in 2024 for the largest income-tax fraud recovery in DC history, and the forensic risk is meaningfully above sector norm even though KPMG has issued an unqualified opinion and no restatement is on the table.

1. The Forensic Verdict

We assign a Forensic Risk Score of 58/100 (Elevated). The accounting policies in the audited financial statements are defensible — ASU 2023-08 fair-value treatment of bitcoin, deferred revenue conventions, and revenue recognition are within GAAP. The forensic risk is concentrated in three places: (1) management's preferred KPIs (BPS, BTC Yield, BTC $ Gain) systematically present accretion that is not reflected in GAAP income or cash flow; (2) the software business is structurally cash-flow negative and cannot service the preferred-stock dividend stack, which is being funded with new common equity; and (3) the controlling founder has two prior, confirmed regulatory matters involving the same legal entity. The single data point that would most change the grade: a restatement, a material weakness opinion from KPMG, or an SEC inquiry tied to the BTC KPI disclosure framework.

Shenanigans scorecard

2. Breeding Ground

The accounting-shenanigans literature treats certain governance and incentive conditions as priors. Strategy Inc. fails three of those tests and passes the rest cleanly. The founder retains effective voting control through Class B supervoting stock, his historical regulatory record includes both an SEC accounting-fraud settlement and a separate confirmed tax-fraud settlement, and the issuer itself was a named defendant in the 2024 tax matter. Against this, the board is majority independent, the audit committee is fully independent and chaired by a CPA-credentialed director, KPMG has audited the company for many years with an unqualified opinion on FY2024 internal controls, and the FY2025 audit committee report confirms no auditor independence concerns.

The 2024 DC settlement is the single most consequential breeding-ground fact. The Office of the Attorney General publicly stated that Saylor "enlisted the assistance of his company, MicroStrategy" to misrepresent his residency, and the company was a named defendant. Saylor and MicroStrategy denied wrongdoing but the $40M payment is the largest income-tax fraud recovery in DC history. The 2000 SEC matter is separately documented in Securities Exchange Act Release No. 43724 (Dec 14, 2000), which describes premature recognition of revenue on unsigned contracts at the same legal entity. These are not allegations; they are settled regulatory facts. They do not predict an accounting restatement today, but they do shift the prior on management's relationship with aggressive presentation.

3. Earnings Quality

Earnings quality at Strategy Inc. has to be analyzed on two tracks: the legacy software franchise, and the bitcoin holdings. The software P&L has been quietly deteriorating for years and is now operating-loss-making even before bitcoin charges, while bitcoin marks dominate reported results and have driven net income from positive $429M (FY23) to negative $1.17B (FY24) to negative $4.23B (FY25). Splitting the two reveals that the underlying business is in retreat even though management's headline narrative pivots all attention toward bitcoin per-share metrics.

Software-only operating income (excluding bitcoin charges)

The blue bars (revenue minus cost of revenue minus selling/general/administrative minus research/development, before any bitcoin-related impairment or unrealized loss) show that what used to be a roughly $50M operating-income software business in FY2020-2021 has slipped to a $40-63M operating loss in FY2024-2025. The red bars overlay the bitcoin-driven distortion. Underlying revenue has been roughly flat at $463-510M for seven years, while cost of revenue rose to $149M in FY25 from $91M in FY20 as customers migrate to cloud subscriptions with higher hosting costs.

Receivables and contract-asset signals

Receivables are the canonical earnings-quality test for a software business. The pattern at MSTR is mild yellow rather than acute red. DSO has drifted upward over seven years, and FY2025 marked the largest single-year jump in accounts receivable (13.5% growth) against revenue growth of 3.0%.

DSO at 148 days is high in absolute terms for a software peer set (peers tend to cluster between 70 and 110), but it has crept up only modestly and there is a plausible business explanation: multi-year cloud subscription contracts under ASC 606 stretch billings over the service period, and MD&A explicitly notes that this transition is depressing cash collection. We grade this yellow rather than red because the gap between AR growth and revenue growth has only widened in one year (FY25 +10.6pp). One more year of similar divergence without an offsetting deferred-revenue jump would warrant escalation.

Restructuring, impairment, and "one-time" item history

Between FY2021 and FY2024 the company recognized approximately $4.06B of cumulative bitcoin impairments under cost-less-impairment accounting. On January 1, 2025, the adoption of ASU 2023-08 reversed those impairments through a cumulative-effect adjustment of $12.75B credited to opening retained earnings — a one-time accounting boost to equity that is mechanical (not earned) and that is then followed by mark-to-market volatility. Management's risk-factor disclosure correctly states that "results for the year ended December 31, 2025 are not, and for future periods will not be, comparable to results from periods prior to our adoption of the guidance." That statement is true, and it is exactly the kind of "comparability break" point that forensic frameworks flag as a moment to read management commentary with extra care.

4. Cash Flow Quality

Operating cash flow has been negative for two consecutive years and has been under 5% of net income on a three-year basis. The cash flow statement, viewed alone, tells the same story the income statement does — the software franchise generates very little cash, and the bitcoin holdings generate none. Where forensic care is required is in the interpretation of how preferred-stock dividends and convertible-note interest get funded.

CFO and FCF versus net income (10-year view)

CFO does not pay the preferred dividend bill

The clearest single picture of cash-flow quality is below: $381M of preferred-stock dividends were paid in FY2025 while operating cash flow was negative $67M. The full distribution cost was funded by sales of class A common stock under the ATM. Management explicitly says it "does not expect cash and cash equivalents generated by our software operations to be sufficient to satisfy our short-term or long-term liquidity needs." This is not hidden; it is just unusual to see this structure presented as if cash flow risk is manageable.

The total annualized preferred dividend run-rate disclosed in the FY25 annual report based on shares outstanding at February 13, 2026 is approximately $32.1M (STRF, quarterly) + $32.4M (STRC, monthly) + $23.0M (STRE, quarterly) + $28.0M (STRK, quarterly) + $35.1M (STRD, quarterly). That implies an annualized obligation north of $750M, against software operations that are losing cash. The S&P Global Ratings B- corporate credit rating assigned in October 2025 reflects this dependency on continuous capital markets access.

Working-capital contribution to CFO

The structural picture: stock-based compensation is the largest single non-cash add-back to CFO ($53-77M annually), without which the software business would have generated meaningfully more negative cash flow over the past four years. Depreciation has risen to $25.5M as cloud-hosting capex grows. Working capital swings (AR, AP) have been modest in dollar terms and not the primary driver. There is no obvious receivables-factoring program in the disclosures, and accounts payable have not stretched abnormally. We do not see evidence of cash-flow shenanigans of the type Schilit catalogs (financing inflows reclassified as CFO, operating outflows reclassified as investing, acquisition-driven CFO boosts). The honest summary is: CFO is what it appears to be, and it is structurally insufficient.

5. Metric Hygiene

The metric framework management has built around the bitcoin treasury is the single most important forensic topic on this tab. It is also the cleanest, in the narrow sense that the company has extensively disclosed the limitations of these metrics — the FY2025 10-K's "Important Information about KPIs" section explicitly says these KPIs "are not, and should not be understood as, financial performance, valuation or liquidity measures" and "should be used only by sophisticated investors who understand their limited purpose and many limitations." But the structural problem remains: management leads with the KPIs in earnings releases, investor decks, and the capital plan, and the KPIs present accretion the GAAP financials do not.

The headline KPI versus the GAAP result

In FY2025, management reports BTC $ Gain of $8.92B as a marker of capital-strategy execution. The income statement reports an unrealized loss on digital assets of $5.40B and a consolidated net loss of $4.23B. These are not contradictory — BTC $ Gain measures change in bitcoin-per-share holdings multiplied by spot price, and it is mathematically independent of the bitcoin price change in the period — but they are easy to confuse, and management's investor communications lean heavily on the favorable framing.

What the KPI excludes

Dilution that is invisible in the per-share KPI

Basic weighted-average shares outstanding grew from 96.8M (FY20) to 277.7M (FY25) — a 2.9x increase in five years. End-of-period basic shares grew from 95.9M to 312.1M (3.3x). Most of this dilution funds bitcoin purchases or preferred-stock dividend payments, both of which sit outside the BPS denominator's "Assumed Diluted Shares Outstanding" treatment if they were funded with convertibles never converted or with non-convertible preferred.

Definition consistency over time

We checked whether KPI definitions have shifted year to year. The BPS denominator definition has changed at least once — the company introduced "Assumed Diluted Shares Outstanding" with bespoke departures from the treasury method, and added STRK Stock to the conversion stack in 2025 with a $1,000 reference price (a tier that becomes dilutive only well above current trading levels). The company has also begun emphasizing "Basic Shares Outstanding" as a parallel denominator that further enhances BPS. Each redefinition has the effect of making BPS look more favorable. None has been hidden; all are disclosed; but the pattern is one-directional.

Balance-sheet metric hygiene

Goodwill and intangibles are effectively zero — there is no acquisition overhang. Net debt to bitcoin holdings is manageable in good price environments. The new and important balance-sheet item is preferred stock with cumulative dividend obligations, which sits structurally senior to common equity and is not netted against bitcoin in the BPS construct.

6. What to Underwrite Next

Strategy Inc.'s accounting policies are within GAAP and KPMG has issued an unqualified opinion. The forensic risk is not "a restatement is coming" — it is "the headline metrics management leads with are systematically more favorable than the underlying economics, the structural cash flow does not cover the cost of the capital stack, and the controlling founder has two confirmed regulatory matters at the same legal entity." Position-sizing matters more than thesis-breaker framing.

Signal that would downgrade the forensic grade (toward High or Critical): KPMG resignation or qualified opinion; SEC inquiry tied to BTC KPI disclosures; reinstatement of full US deferred-tax-asset valuation allowance; receivable factoring or supplier-finance program appearing without disclosure; AR-to-revenue gap widening for a second consecutive year; departure of CFO Andrew Kang without a planned successor.

Signal that would upgrade the grade (toward Watch): Software operating cash flow returning to positive territory on a four-quarter basis; preferred dividend coverage from operations exceeding 0.5x; explicit board-approved policy capping ATM-funded dividend payments; KPI architecture simplified or supplemented with a clear GAAP reconciliation table; settlement of all securities lawsuits with no follow-on filings.

The decision-grade close

The accounting risk at Strategy Inc. is not a footnote and it is not a thesis breaker. It is a valuation haircut and a position-sizing limiter. The audited GAAP financial statements are reliable; the issue is that the metrics the company asks investors to use as a substitute for those financial statements are structurally more favorable than the financial statements themselves. A long position should not pay full economic value for BTC $ Gain or BPS accretion that has not flowed to GAAP net income, GAAP equity attributable to common, or operating cash flow. A short position should not assume that a restatement is the catalyst — the catalysts are bitcoin price (the dominant variable), preferred-dividend coverage stress, and capital-markets access. The controlling founder's documented regulatory history justifies a higher discount rate or governance-overlay haircut, especially when reading favorable management commentary on an unaudited interim basis.