Hook: On August 14, 2026, a peculiar data point emerged from the financial engineering lab of Strategy (formerly MicroStrategy). STRK, one of its four preferred stock series, had returned -27% over the past year. STRC, the floating-rate preferred, had returned +9%. Bitcoin had dropped 47%. MSTR common stock? Down 75%. This is not random variance. It is a deliberate decomposition of risk—and a warning about the limits of financial engineering in a bear market.
Context: Strategy began as a software company. In 2020, it shifted to a Bitcoin treasury strategy, borrowing at low rates to buy BTC. By 2024, the model had evolved: instead of issuing convertible bonds, it began issuing preferred stock. Four series emerged: STRC (12% annual coupon, floating rate to maintain $100 par), STRD, STRF, and STRK (convertible into 0.1 shares of MSTR). The total preferred stack reached $15 billion. The narrative was simple: convert Bitcoin’s volatility into a fixed-income-like instrument for yield-seeking investors, while keeping the upside for common shareholders. But the bear market of 2025-2026 has exposed the structural flaws in this design.
Core:
- The Floating Rate Illusion: STRC’s mechanism allows the company to adjust the rate to keep the price near $100 par. This summer, STRC broke par. Why? Market participants began pricing in the risk that the company may not sustain the 12% dividend if Bitcoin continues to decline. The rate adjustment is a tool, not a guarantee. In my 2020 Curve stress test, I simulated a 15% depeg. The invariant failed when liquidity evaporated. Here, the invariant is the company’s cash flow. When that liquidity is questioned, the par price becomes a memory.
- The Backstop Price Gap: The company has not publicly disclosed the “backstop price” for each series—the Bitcoin price at which the preferred stock’s principal is at risk. Based on the leverage ratio (approximately 1.6x from the 75% common stock drop vs. 47% BTC drop), and the fact that the company sold 1,638 BTC in a week after buying 37, I estimate the backstop for STRC is around $25,000–$30,000. Below that, the dividend obligation exceeds the company’s ability to generate cash from operations or BTC sales without destroying the asset base. This is the same structural flaw I found in the 0x whitepaper in 2017: a hidden assumption about liquidity continuity. Here, the assumption is that Bitcoin will never stay below $30,000 for a prolonged period.
- The Leverage Shock: The common stock’s 75% decline is not a market overreaction—it is a mathematical consequence of the preferred stack. The company has effectively created a leveraged position: the preferred shares absorb the first losses, but the common equity bears the residual. In a bear market, the leverage factor increases as the equity base shrinks. If BTC drops another 30%, MSTR could fall 90%+. This is the same dynamic I simulated in the Curve three-pool stress test: once the invariant is breached, the system collapses faster than the underlying asset.
- The Dividend Burden: STRC pays 12% annually on $100 par. The entire preferred stack requires annual dividends of roughly $1.8 billion (based on $15 billion at 12% average). Where does this cash come from? The company’s software business generates minimal free cash flow. They have been selling Bitcoin to fund operations. The net selling of 1,638 BTC in a week is a clear signal: the company is now a net seller, not a buyer. This creates a negative feedback loop: selling BTC depresses price, increases the risk of further dividend cuts, and forces more selling. This is the “death spiral” I documented in the Terra Luna collapse. The mechanism is different—here it’s not algorithmic, but the incentives are the same: when the asset price falls, the structure forces liquidation.
- Selective Disclosure: Michael Saylor’s public communications show a chart of preferred stock returns vs. Bitcoin, conveniently omitting the 75% common stock decline. This is not a simple oversight. It is a deliberate framing to attract yield-seeking investors while hiding the cost to the equity base. In my 2021 BAYC audit, I found that the metadata update logic had no ownership transfer restrictions—a minor flaw that became a centralization risk. Here, the omission is a disclosure risk. If the SEC examines the consistency of these communications, the company could face regulatory scrutiny. Ownership is an illusion without immutable proof.
Contrarian: The bulls are not entirely wrong. The preferred stocks did provide downside protection. STRC’s +9% return in a -47% Bitcoin market is a genuine achievement in financial engineering. The volatility transformation works—for the preferred holders. The structure is not a Ponzi scheme; it is a legitimate risk decomposition. The problem is that the cost is externalized to common shareholders and the company’s long-term viability. The bulls argue that this is a feature, not a bug: the preferred stock absorbs volatility, allowing the company to hold its Bitcoin. But the data shows the company is now selling. The “accumulate forever” thesis is broken. The real insight is that the financial engineering is a zero-sum game within the corporate structure. The preferred holders’ gain is the common holders’ loss. And when the market turns bearish, the cost becomes systemic.
Takeaway: The next phase will test the backstop. If Bitcoin continues to trade below $40,000, the company will face a choice: cut dividends, sell more Bitcoin, or issue new securities. Each option accelerates the negative feedback loop. The question is not whether the structure is clever—it is. The question is whether it can survive a multi-year bear market. History suggests that leverage in a bear market is a one-way door. The preferred stocks may hold their value for a while, but the common stock is the canary in the coal mine. Trace the exit liquidity: the common shareholders are the exit liquidity for the preferred. When the music stops, the promises expire. Code executes, promises expire. Strategy’s financial engineering is a brilliant experiment, but experiments have a failure rate. The data is clear: the model is under stress, and the stress is directional.
