Six loans, one sponsor, plus a redemption queue I never priced

The SPV thing is what stops me cold. If every loan is to a differently named LLC in a different state, how would a normal person outside the fund ever see that six of them are the same guy? Is that something you can actually ask for and get, or did you only find out because they told you after it broke?

@solder you can ask. Whether you get it depends on the manager. I walked from a fund last spring for close to the same reason and I found it the ugly way, by pulling the property addresses off the loan schedule and running them through county records. Four of them shared a mailing address for the borrower entity. Bear in mind a fair number of states don't publish sale prices at all, so that trick works better in some places than others.

Canton, the part that would have kept me up isn't the concentration. It's that the gate and the concentration hit in the same quarter. Everybody wants out at once precisely because the loans went bad, so the two risks are the same risk wearing different hats.

If the loans are secured by real property, why does NAV drop at all? The building is still there. And what does non-accrual actually mean, that they stopped paying or that the fund stopped counting it?

@vellum non-accrual means the fund stopped recognizing interest income on that loan because collection is no longer reasonably expected. The borrower may still be paying something. The accounting just stops assuming it.

Canton, the clause I'd have gone looking for is the interaction between the 5 percent cap and the suspension power. In a lot of these documents the cap is measured on NAV at the start of the quarter, so a falling NAV shrinks the size of the queue payment in the exact quarter you want it largest. Then the suspension right usually sits with the manager or an affiliated board rather than an independent one, and the trigger language is often something like "where orderly disposition is not practicable," which is broad enough to cover a lot.

The other thing worth pulling is whether any LP has a side letter with preferential redemption or a most-favored-nation clause. Sometimes the fund will tell you if you ask directly in writing. Sometimes the answer to "are there side letters" is the whole diligence.

The diagnosis here depends on one assumption: whether the four non-accrual loans were construction loans funded with interest reserves. If they were, the seven quarters of clean payment history told you very little. A construction loan with a capitalized interest reserve pays itself out of the loan proceeds until the reserve is exhausted. Performance during the draw period reflects the reserve balance rather than the borrower's ability to service debt. The signal only arrives at reserve burn-off or at maturity, which is why construction-heavy books can look pristine right up until they don't.

On concentration, borrower-level exposure is a distinct measure from property type and geography, and the three move independently. A book that is 90 percent multifamily across eleven states can still be 20 percent to one sponsor. Managers who underwrite well generally track it and will produce the table. Ones who resist the question are telling you something.

On valuation, ask which policy the fund uses. Loans held at amortized cost move only when the manager takes a specific reserve, so NAV can look stable while credit deteriorates. Fair-value books mark sooner and look more volatile for the same underlying facts. Neither is wrong, but comparing net returns across two funds using different policies is comparing different things.

On liquidity, an open-end debt fund holding illiquid loans has a structural mismatch that redemption caps exist to manage. That is the design working as intended, not a failure. The mistake is treating quarterly liquidity as a feature you can rely on in the quarter you most want it.

What I'd add to your list: get the fund's history of actual redemption fulfillment by quarter, in writing, going back through 2020. Stated terms and honored terms are different data sets. Have counsel read the redemption and suspension provisions before you subscribe.

Two documents nobody reads and both would have flagged this. The valuation policy in the offering memorandum, which tells you when a loan gets reserved and who signs off. And the subsequent events footnote in the last audited financials, which is where extensions and forbearances tend to appear first because they happened after the balance sheet date and before the auditor signed.

I've caught two problem books in that footnote. It's usually four sentences and it's usually the most honest paragraph in the document.

Canton, did the audit for FY2023 say anything about extension activity? I'd bet those construction loans had been extended at least once before the non-accrual.

I read the cause differently to most of this thread. Concentration made the hole bigger. It didn't dig it. Six separate sponsors on the same four construction loans and you'd still have four non-accruals, just spread around.

The underwriting failure is the exit. Construction lending is a bet on a takeout, either a sale or a permanent loan at completion, and both of those got repriced hard between 2022 origination and 2024 completion. If the fund underwrote exit cap rates and perm debt spreads at 2022 levels, every construction loan in that vintage is impaired regardless of who the borrower is. Canton, ask them what exit assumption sat in the credit memo on those four. If it's a 2022 cap rate with no stress case, the concentration table wouldn't have saved you.

Even so, your second lesson stands unqualified. Modeling the gate as base case is the right correction and it's the one most LPs never make.

Losing 4 percent of NAV is survivable. Losing access to 78 percent of your money for a year while you watch is the part that would have wrecked me.