Dynex Capital (DX) posted earnings of $0.38 per share for the September quarter, up 52.0% from a year ago, on revenue of $332.5 million, up 122.1%. Those figures show how fast the REIT has scaled its balance sheet, but they are not what mattered this quarter. Mark-to-market losses swamped them. Book value per common share fell $1.99, or 15.4%, to $10.91. Total economic return was a negative 11.5% of beginning book value, and the company reported a GAAP net loss of $1.34 per share, or a comprehensive loss of $1.49, compared with net income of $0.80 in the June quarter. That gap is the central tension. Earnings power is growing, while the capital base those earnings sit on just took its worst hit in at least a year.

The $0.38 is earnings available for distribution, a non-GAAP measure. It improved from $0.36 in the second quarter for three reasons:

• **Larger asset base.** The average balance of interest-earning assets grew 10%.
• **Higher TBA drop income.** It rose to $8.4 million from $5.2 million.
• **Lower operating expenses.** They fell about $4 million because of reduced stock-based compensation.

Underlying profitability per dollar invested was essentially flat. Net interest spread inched up to 1.18% from 1.16%, with asset yields at 5.00% and funding costs at 3.82%. Period-end repo rates climbed to 3.91% from 3.79%, so financing costs are not moving in Dynex's favor.

The dividend math deserves scrutiny. Dynex again declared $0.51 per share, well above $0.38 of distributable earnings. Management's call framing rested on preserved earnings power rather than an explicit coverage figure. There is a partial bridge. The company carries $625.6 million of deferred tax hedge gains that will flow into REIT taxable income over the coming years, roughly $96-98 million annually in 2026 and 2027. Those gains create distribution requirements, but they do not show up in EAD. Investors should not mistake them for recurring spread income.

The call gave a candid picture of what went wrong:

• **Spreads blew past the forecast.** The 10-year Treasury broke 5% and mortgage spreads widened to roughly 162 basis points. Management had earlier pointed to a 100-120 basis point equilibrium.
• **The backstop thesis fell short.** The GSE-support argument it leaned on in prior quarters did less to cap widening than suggested.
• **Hedges helped but could not keep up.** They worked roughly as intended. About $1.17 billion of gains on swaps and Treasury futures offset most of the $1.50 billion markdown on MBS. But the spread move was beyond what rate hedges can address.
• **Leverage rose without being chosen.** It reached 9.0 times equity, above the 7.5-8.5 times comfort range cited last quarter, mainly because book value shrank rather than by deliberate choice.
• **Liquidity thinned.** Cash and unpledged securities slipped to $1.3 billion, or 42% of equity, from $1.6 billion, or 51%.
• **Messaging turned defensive.** The earlier flywheel narrative gave way to the line that the portfolio "operated as designed."

The bull case is that the damage has reset forward returns. Management now quotes static returns on equity of roughly 20% in lower coupons and 25% in current coupons at about 10 times leverage. That compares with mid-to-high teens in the first quarter. The company's sensitivity analysis also supports the case:

• **Rate risk.** A 50 basis point rate shock would cost only about 0.5% of common equity, helped by a mortgage universe with little refinancing incentive.
• **Spread upside.** Twenty basis points of spread tightening would add roughly 11.5% to book value.

Dynex leaned in rather than retreating. It deployed $384 million of new capital into $3.2 billion of MBS purchases. That included its first preferred offering since 2020, a $116 million 9.375% Series D. The deal diversifies funding but adds a relatively expensive fixed claim ahead of common holders. Management also pointed to GSE value buying, possible foreign demand, and net purchases since quarter-end. Book value was $10.95 in mid-October, essentially unchanged from September 30.

The risk list is equally concrete:

• **Higher-for-longer pressure on rates.** Management cited inflation tied to AI capital spending, fiscal and political pressure on yields, and higher implied volatility.
• **Selling pressure in mortgages.** It flagged continued floater and convexity selling.
• **Little cushion at current leverage.** Running near 9 times leverage with thinner liquidity leaves less room if spreads widen again.

The ATM program also expanded the share count to 258.4 million from 237.4 million. Accretion now depends heavily on where the stock trades relative to a much lower book value.

Investor sentiment and current price data were not available for this review. One useful reference point remains. Shares opened at $13.38 after the July report, modestly above the $12.90 book value at the time, and the 200-day moving average stands at $13.07. With book now near $10.91, any price near those levels would imply a materially richer premium to book than the stock carried three months ago. That would make the valuation more dependent on management's forward-return thesis playing out.

The bottom line is that Dynex delivered sequential growth in distributable earnings while absorbing a severe book value drawdown. Management chose to press its bet with more leverage and fresh capital rather than retrench. The 20-25% forward returns it quotes are real if spreads stabilize or tighten. But that is a projection, not a result. Meanwhile, $0.38 of EAD against a $0.51 dividend and leverage above its own comfort range give skeptics legitimate ammunition until spread behavior turns.