Three Marks, One Asset

The $1 trillion private credit book that just failed its easiest test. Standalone private credit analysis — Wednesday, August 12, 2026. All prices as of ~12:15 p.m. ET.

1. Why today is the tell, not just another red day

For eighteen months the standard defence of private credit has been mechanical: these are floating-rate loans, the problem is that SOFR above 4% crushed middle-market interest coverage, and the fix is lower short rates. Every sell-side deck says some version of it.

This morning the market got the first clean instalment of that fix. Per the BLS release, headline CPI rose 0.1% m/m to 3.4% y/y and core rose 0.2% m/m to 2.5% y/y — both a tenth lower than June, both exactly in line. Treasury yields fell across the board. CME FedWatch cut September hike odds to 42%. Morgan Stanley's Ellen Zentner: "In-line inflation will keep the 'no need to hike rates' narrative intact."

Here is what each asset class did with that information:

Public credit is fine. Bank credit is fine. Only the vehicles that hold privately marked credit are down. Two days ago these same six alt managers were the best-performing group in the entire market, up 6–7% on the Nvidia $500 billion compute-financing headline. They have now given all of it back inside 48 hours while the index made a new high.

2. The primary source almost nobody has read

Seven days ago, on August 5, three economists at the Federal Reserve Bank of Boston published Current Policy Perspectives 26-6, "Early Warning Signals in Private Credit? What BDC Portfolios Reveal about Emerging Risks" (Fillat, Shen and Wang; full PDF here). It got essentially no press. It is the single best dataset in existence on this asset class, because BDCs are the only private credit vehicles legally required to publish a complete Schedule of Investments.

The authors harmonised 168 BDCs and roughly 890,000 company-quarter loan observations from 10-Q and 10-K filings. Read the sections on PIK usage, on spreads, and on fair-value ratios. Four findings matter:

Two further items from the paper that go straight into the trade:

• Software is ~20% of the median BDC portfolio by loan count, with an interquartile range of 15–30% and "some BDCs [with] more than one-third of their lending in technology sectors." The authors explicitly flag AI disruption of software business models as a credit risk, not just an equity risk.
• The April 2026 SLOOS showed large and regional banks tightening standards on business credit intermediaries and PE funds across five dimensions simultaneously — maximum loan size, maturity, risk premiums, covenants and collateral — while demand from BDCs strengthened. Tightening supply into rising demand is a funding-cost event with a lag.

3. The number nobody is talking about: 48.5 cents

Take the Boston Fed's industry mark of ~100 cents on cost and hold it next to an actual filing. FS KKR Capital (FSK) is the second-largest listed BDC. Its Q2 2026 release and balance sheet (filed August 6) disclose the following, and the derived figures are mine:

So there are three marks on the same pool of middle-market loans. The industry says ~100 cents. FSK's own valuation committee says 90.6 cents. The public equity market says 66.4 cents. End to end, that is a 34-point spread on an asset class that is supposed to be senior secured.

One of those three numbers is going to move a very long way. It will not be the one set by people whose fees are calculated off it.

The signal inside FSK's capital structure

Buried in the same release: a KKR subsidiary completed a $150M tender for FSK common, FSK issued $150M of convertible preferred to a KKR subsidiary (6,000,000 shares, $25 liquidation preference), and FSK began a $300M buyback, purchasing 3,726,153 shares at a weighted average of $10.73.

Do the arithmetic the release does not do for you. The Q3 preferred dividend of $0.315972 annualises to $1.2639 on $25 — a 5.06% coupon, cheaper than FSK's own 5.49% weighted average cost of debt, while the common yields 14.4%. And the buyback, executed 41% below NAV, adds roughly $0.10/share of NAV accretion — against $0.53/share of markdowns in the same quarter. The manager is genuinely subsidising this vehicle. It is losing the race 5-to-1.

4. Where consensus is wrong

The consensus fear is "private credit blows up the banks." That fear is quantitatively wrong, and it is costing people money in the wrong short.

The Boston Fed sizes it precisely: total committed bank credit to BDCs grew from ~$10B in 2013 to more than $50B by 2025, with ~$35B drawn. That is less than 2% of large-bank Tier 1 capital, and the claims are predominantly senior secured — banks are first in line. The authors' conclusion is unambiguous: losses in a severe stress scenario "would not be large enough to threaten bank solvency," and "the direct transmission of private credit stress to banking-system stability seems to be limited."

The tape agrees. KRE was +0.68% today while APO was −2.34%. If this were a bank-contagion story those two numbers would have the same sign.

The real transmission runs through three channels nobody is short:

1. Manager fee earnings. Perpetual non-traded BDC AUM is the highest-margin dollar an alt manager owns — permanent, retail, full fee load. It is also the AUM that is now shrinking.
2. The wealth channel's plumbing. Gates are not a liquidity inconvenience; they are a distribution kill-switch. A gated product loses shelf space at the wirehouses for years.
3. Middle-market credit supply. Gated vehicles cannot deploy. Remove the marginal bid and spreads should widen — yet the Boston Fed measures them compressing 100bp. Both cannot be true.

5. The redemption ledger

Fitch's July 7 note is titled, plainly, "Elevated Redemptions Persist for Perpetually Non-Traded BDCs." The individual disclosures:

Non-traded BDCs handed back more capital than they raised in Q1 2026 — the first net-outflow quarter in the product's history. Raymond James' global head of private capital advisory, Sunaina Sinha Haldea, put it to CNBC as well as anyone has:

"We're discovering in real time that you can't offer near-daily liquidity on genuinely illiquid assets without eventually testing the plumbing, and 2026 is the year those structures get rewritten."

The counterpoint, from Oaktree's Danielle Poli on the same day, is fair and worth holding: the retail wealth channel is under a quarter of the private credit market, and institutions are adding into scarcer capital. I think she is right about the asset and wrong about the equities. The listed managers' incremental earnings growth was sold to shareholders as a wealth-channel story. That is the part that is breaking.

6. What the default rate actually contains

Fitch's U.S. Private Credit Default Rate has sat at a record 6.0% since April 2026 (May release), after a 9.2% full-year 2025 print. But the composition is the story, and it cuts both ways. On the February vintage of the series (72 borrowers, 87 default events over twelve months, component breakdown here):

• 55% — interest deferrals or switches to PIK
• 31% — stressed maturity extensions
• 6% — uncured payment defaults
• 8% — true breaks (bankruptcy, liquidation, debt-for-equity, restructuring)

The bull reading: only 14% of "defaults" are hard events. Recoveries should be far better than a 6% headline implies.

The bear reading, which I hold: 86% of the default events are the lender choosing to not call a default — extending the maturity or accepting paper instead of cash. That is the same behaviour the Boston Fed independently measures from the other direction as PIK going 6% → 10%. Two datasets, two methodologies, one conclusion: the losses are being deferred, not avoided. And the deferral is being booked as income.

7. FSK vs. OBDC: the same discount is not the same trade

Both are cheap. Only one has a book you can underwrite. Blue Owl Capital Corp (OBDC) reported Q2 on August 5:

OBDC is not clean. It cut its base dividend 16% (from $0.37 to $0.31) and its NAV has fallen four consecutive quarters ($14.89 → $14.81 → $14.41 → $14.26). Its advisor faces a Section 36(b) derivative complaint filed with the SEC alleging that Blue Owl Credit Advisors "systematically inflated the value of the fund's assets" to extract excessive advisory fees. And in February the Blue Owl BDCs sold $1.4 billion of loans to four institutions at 99.7% of par — a transaction the manager cites as validation of its marks, and which the public market rejects by 18 points of discount every single day.

FSK's own securities class action complaint (S.D.N.Y., filed July 15, 2026) notes the stock traded at a 45.9% discount to reported NAV as of March 31. That is the relevant downside reference point, not zero.

8. Scenarios into year-end

Bull — 25%

Trigger (measurable): September FOMC hike odds fall below 25% on CME FedWatch and the Q3 BDC prints show sector non-accruals down for a second consecutive quarter and the Boston Fed's PIK share stops rising. The 6.0% Fitch PCDR ticks to 5.5% or below.
Outcome: average BDC discount compresses from ~20% toward ~10%. OBDC $13.20+, BIZD +12–15%, alt managers re-rate 15–20%. The 86%-deferrals-not-defaults bull reading proves correct and recoveries surprise.

Base — 55%

Trigger: marks grind down 1–2% per quarter; at least one more large BDC dividend cut lands in Q3 or Q4; gates stay at 5% through the October redemption window; PIK share prints above 10% in the next Boston Fed update.
Outcome: OBDC ranges $11.00–12.50. FSK drifts to $10.50–11.50. Fee-centric managers (BX, BAM) outperform balance-sheet/wealth-channel managers (APO, ARES, OWL) by 500–800bp over three months. No systemic event.

Bear — 20%

Trigger: any one of — a top-five manager marks a flagship book down more than 5% in a single quarter; a gate is extended beyond one quarter or a vehicle suspends entirely; the §36(b) case survives a motion to dismiss with damaging valuation discovery; or the Fitch PCDR breaks 7.0%.
Outcome: BIZD −15–20%. FSK to $9. OWL to $9. APO to $118. This is where the 34-point mark gap closes violently and the sequence is filings → forced sales → marks, in that order.

9. The trades

10. Scorecard on prior calls — including the losers

11. What to do tomorrow

1. Close the VIXY hedge. The binary it insured resolved this morning. Do not let a resolved hedge decay.
2. Put on the OBDC / FSK pair at market, dollar-neutral. This is the highest-conviction idea in the note and the one least dependent on the direction of anything.
3. Add to BX in $138–147 on any continued group weakness. Long the fee-collector, short the mark-taker.
4. Do not short regional banks on private-credit headlines. Buy them into that panic.
5. Set an alert on BIZD at $12.40. That is the sector entry, not today.
6. Diarise the Q3 prints (early November) and watch exactly three numbers per BDC: non-accruals at amortized cost (not fair value), fair value as a percentage of amortized cost, and whether NII covers the dividend before any fee waiver. Those three lines are the whole thesis.
7. Watch the §36(b) docket. A denied motion to dismiss is the single highest-impact catalyst in this asset class and it is not on anyone's earnings calendar.

The thesis in one paragraph

Private credit is priced in three places at once and the prices are 34 points apart: the industry marks its book at roughly par, FSK marks its own book at 90.6 cents with its non-accruals at 48.5 cents, and the public equity market marks FSK at 66.4 cents. The Boston Fed's August 5 study of 890,000 BDC loan observations shows why the industry mark is not credible — PIK usage up 67% since 2022 while spreads compressed 100bp, which is lenders accepting less compensation for visibly worse cash flows — and it shows that the disclosed marks predict subsequent equity returns, making this a factor rather than a story. Today the market delivered the private credit bull case exactly what it asked for, a tame CPI and lower yields, and the entire complex sold it while tech, banks and high yield rallied. That is the market conceding this is no longer a rates problem. The trade is therefore not "short private credit" — the bank-contagion version of that short is quantitatively wrong at less than 2% of Tier 1 capital — but long the fee-collectors and the balance sheets that cover their dividends unsubsidised, short the ones being kept alive by fee waivers and liquidation. Long OBDC, long BX, short FSK. The marks will close the gap on a court calendar, not a macro one, which is precisely why it will happen later and faster than anyone is positioned for.