Question
Can a fund’s own marks reveal stress before the non-accrual rate does?
Method
Can a fund’s own marks reveal stress before the non-accrual rate does?
The method behind a private-credit data investigation, published before the result.
Overview
Can a fund’s own marks reveal stress before the non-accrual rate does?
The method is complete. The result is not locked — a recompute is open.
Listed BDCs · Form 10-Q, quarter ended March 2026 · debt only · anonymised by design.
No number is published here. This is the method, not the result.
Domain logic, decisions and verification mine. Implementation built by directing AI coding tools.
The question
Funds publish a non-accrual rate — the official distress number. Their filings also carry a position-by-position schedule: every loan, what it cost, and what the fund itself marks it at today.
The claim is narrow: headline non-accrual rates may understate stress that funds’ own marks already reflect. Not that anything is concealed — the marks are published, in the same document, by the same fund. If the mark moves before the status does, a lender watching the headline rate is watching the slower signal.
The lens
Accruing, and marked at or near carrying value.
Still accruing. Performing on paper, impaired in the fund’s own pricing.
The fund’s own designation. Already caught by traditional monitoring.
The control
A markdown on a fixed coupon can be a rate artifact. A markdown on a floating coupon is a statement about the borrower. This is the control the method sits on.
The decisions
Fair value already contains the markdown being pointed at, which softens the very thing under examination. One basis throughout means the population never switches mid-calculation.
Measuring against the floating-rate book alone excludes exactly the loans that matter where distress sits in fixed-rate tranches. The control stays where the inference is.
Some funds report distress figures that include preferred equity. Matching them would mean inferring which stakes were impaired — and an inferred number has no business in a reconciliation.
One fund’s schedule tags a different amount as distressed than its own narrative reports. The schedule wins: it is the primary source, and the narrative is derived from it.
One fund publishes on only one of the two bases, so on the basis used here there is nothing to verify against. It carries a standing flag saying so.
Take the smaller, fully verified number over the bigger, unverified one, every time.
A structured tag is absent for some floating-rate loans. Correcting it moves loans in on both sides.
Pooling and averaging give different answers where a denominator is small.
Some funds fall below the data-quality threshold; one runs on a documented fallback.
Publication standard
Every figure that reaches a published conclusion carries all four together. Until the recompute closes, the method is finished enough to examine and the result is not.
If you price this kind of risk for a living and something here does not hold — a control, a boundary, a call I got wrong — I would rather hear it before the number publishes than after.