Agency at 7x or commercial bridge at 2x. Which risk am I paid for?

Run economic return on both for four years and stop looking at yield. Name A: book from 15 to 9.20 is minus 5.80, plus roughly what, 1.30 a year in dividends, call it 5.20 of income. Net minus 0.60 per share on a starting book of 15. That's slightly negative over four years while advertising 14 percent. Name B with flat book and 10 percent paid is close to 10 percent compounded, assuming the marks on those risk rated 4s and 5s are honest, which is the whole assumption. Ask what the reserve percentage is against those loans and how it moved.

Seventeen percent in the bottom two risk buckets is the number I'd be pulling on. Ask three things of the filing: what the weighted average loan to value was at origination on those specific loans, whether the sponsors have funded interest reserves that are now exhausted, and how many of them have been modified or extended in the past year. Extensions are where commercial bridge books hide their problems, because a loan that got twelve more months stays current and stays out of nonaccrual. Two on nonaccrual out of a book that size is not alarming by itself. Two on nonaccrual with a dozen recently extended is a different picture.

Borrowing for a month to lend for five years is the same trick a sub pulls when he floats payroll on a credit card against a slow paying GC. Works fine until one link stalls.

The 0.78 times book on Name A isn't a bargain, it's the market telling you it doesn't believe the book. Discounts like that persist for years in this sector and people keep buying them expecting a snap back to par. Meanwhile Name B at 0.95 is priced like the loan book is fine, and canton is right that the reserve assumption is doing all the work. You're picking between a discount that might be correct and a premium that might be unearned.

Before the split question, what's the alternative use of the money. If it's a down payment on something you'd operate, the comparison isn't A versus B, it's both of these versus a deal you control. I ask because I keep talking myself into paper when what I actually want is an asset I can touch, and I don't think that's a good reason to buy either.

Genuinely asking, what does repo mean here. I've read the word four times in this thread and I've been assuming it's something to do with repossession, which I suspect is wrong.

rigging, it's a repurchase agreement. The company sells its bonds for cash and agrees to buy them back a few weeks later at a slightly higher price. It's a short term loan with the bonds as collateral, and the difference in price is the interest.

The construction slug in Name B is the part I'd stare at. On the renovation side I've watched a budget go from fine to 30 percent over in about ten weeks once the structure opens up, and a lender holding a construction loan on that project has a borrower who needs more money before anything is finished. The lender either funds the overrun or owns a half built building. Ask how much of that construction exposure is vertical versus already stabilized and how many draws are behind schedule. A stalled draw schedule is the earliest warning you get.

Pull the dividend history on both going back through the last rate cycle. If Name A has cut twice and reverse split once, that 14 percent is a number attached to a much smaller base than it looks. I want income that shows up, and a payout that has been reset three times is not income, it's a series of announcements.

What happens to Name A when rates fall and everyone refinances. I read that prepayment is a real risk for the agency ones because the high yielding mortgages disappear and get replaced with lower yielding ones, and the company paid a premium for those bonds in the first place. Does that show up as a book value hit or as lower earnings, or both.

juniper, both, and that's the part that makes them awkward to hold alongside stocks. On the split question, plumb, splitting between an agency book and a commercial credit book isn't two versions of the same bet. They diverge sharply on a steepening curve. What they do share is that both get hit when funding markets get scared, so you're diversified against rates and credit while staying fully exposed to liquidity stress.

Ten years out, the question is whether the manager compounds book value or spends it. Everything else in this thread is timing. I'd take the four year economic return canton calculated, do it again for eight years and for twelve if the company has been around that long, and see whether the shape repeats. A manager who gave back capital through two different rate environments is telling you something about how they run the balance sheet. That's a longer test than any spread number you can pull today.

On the multifamily bridge loans, I'd want to know what markets those assets sit in. We manage in a couple of metros where new supply landed hard and rent growth went negative for six quarters, and every value add bridge deal underwritten on rent bumps in those submarkets is now short of its plan. The loan can be current and the business plan can still be dead. That gap is where the risk rated 4s come from.

trellis just described the actual due diligence. Geography of the loan book beats any portfolio level ratio you'll find in the summary tables.

Reading this back, your original framing had the answer in it. You said A fails slowly and B fails in one quarter. If you can't stomach opening a filing and finding a 40 million reserve you didn't see coming, size B smaller and take the slow bleed you can monitor. If you'd rather have a bad quarter than a bad decade, weight the other way. Both of those are decisions about you rather than about the spread environment, and I don't think there's a number that resolves it.