The repo maturity table in this 10-K is doing more work than the dividend

Do the arithmetic on the spread before you worry about the plumbing. Take the weighted average asset yield, subtract the average cost of funds including the hedge cost, and you get net interest spread per dollar of asset. Multiply the spread by the 6.8 dollars of borrowed money per dollar of equity, then add the yield you earn on your own equity dollar, and you get gross return on equity before management fees and operating costs. If the asset yield is 5.1 and all in funding is 4.4, that's 70 basis points on 6.8 dollars of borrowings, about 4.8 percent, plus the 5.1 on the equity dollar, so about 9.9 percent on equity before costs. That's still under a 12 percent payout, and the management fee and operating costs come out of it, which tells you either my inputs are stale or part of that dividend is capital coming back. Post the two yields and I'll redo it.

Notional on the hedge book tells you the size of the position and nothing about how well it matches. What you want is the duration gap after hedges, which most of them disclose in the interest rate sensitivity table, along with the estimated change in book value for a 50 and 100 basis point shift. If the gap is near zero they're hedging parallel moves reasonably well. Neither of those helps with spread widening, where the mortgages cheapen against treasuries and the hedge doesn't move with them. That's the risk that has actually eaten book value in this sector.

I've been trying to read the same kind of table for two weeks and this is the first explanation of repo that stuck. The part I still can't square is the margin call. If the bonds drop and the lender wants more collateral, where does the cash come from if the company is already levered near seven times. Do they hold a cash buffer for exactly that, and is it disclosed anywhere I'd find it.