The Exit Door Repriced

Private credit just printed three real transaction prices in five weeks — 15–30% below NAV for retail BDC shares, roughly 90–95% of NAV for institutional fund stakes. Every one of them sits below the 100 cents the funds are still reporting. That is not a discount. That is an unrecognized loss with a settlement date.

Daily Analysts · Standalone private credit deep dive · Tuesday, July 28, 2026 · price data timestamped 16:00 UTC / 12:00 ET intraday

1. What actually happened — three prints in five weeks

Print #1 — June 16/17: the most patient capital in the world became a seller

Singapore's GIC hired Evercore to sell up to $2 billion of private credit fund stakes (Benzinga, June 17). GIC has a perpetual horizon and no redemption liability. It does not need liquidity. When an investor with no liquidity need sells anyway, the sale is a view on forward return, not a cash-management exercise. Private credit stake sales had already hit ~$20B in the period.

Print #2 — July 17: a third party bid 70–85 cents on retail paper

Per the dated headline feed on the Yardeni Private Credit Monitor, Cox Capital tendered for Apollo, Ares and HPS non-traded BDC shares at 15–30% discounts to NAV — in the same quarter that redemptions across those vehicles averaged 10.3% of shares and were capped at 5%. This is the most important number in the piece, and almost nobody has traded it. A tender at 70–85 cents on the dollar is not a valuation opinion; it is a bid that a gated shareholder can actually hit today. It establishes the clearing price for the marginal seller of non-traded private credit — and it is 15 to 30 points below the NAV those same shareholders see on their statements.

Print #3 — July 27: the GP itself organized the exit

Ares Management is in discussions with credit-secondaries buyers to sell $3.4 billion (€3 billion) of limited partner stakes in the fourth vintage of Ares Capital Europe — potentially one of the largest private credit sales on record (Benzinga, July 27; talks are ongoing and no deal is guaranteed). Ares' own published research on the asset class puts credit-secondaries volume at $15B in 2024 — a 5x increase from 2019 — with advisors projecting $28B in 2026 and above $50B by 2030, and typical pricing framed in discount bands of 5%, 10% and 15% to NAV. Dave Schwartz, Ares' head of credit secondaries, is explicit about the motive: sellers are using the market "to access liquidity in order to prudently manage downside risk against an uncertain market backdrop" (Ares).

2. The liquidity data underneath the prints

The prints did not appear from nowhere. The redemption machinery broke first. From the dated headline record on the Yardeni Private Credit Monitor:

Quantify the mechanism, because this is where most commentary stops. $15.6B requested and 38% honored means roughly $9.7B of investors asked to leave in Q2 and were not allowed to. That queue does not evaporate; it rolls into Q3 and stacks on top of new requests. A rolling, gated queue is precisely the condition under which the secondary market — not the fund's own NAV committee — sets the marginal clearing price. Two consequences follow mechanically:

1. A gated fund cannot originate. Cash that would fund new loans is reserved for the repurchase facility. Tens of billions of origination capacity leaves the market, new-issue direct lending spreads widen, and the lenders with permanent capital — listed BDCs and insurance-funded balance sheets — get to price loans for the first time since 2021. This is a structural advantage, not a cyclical one.
2. Fee-related earnings are levered to NAV, not to defaults. Management fees are charged on gross assets or NAV. A 10-point NAV haircut is a permanent 10% revenue reduction on that vehicle, before any performance fee impact. This is why the repricing hits Blue Owl, Apollo and Ares equity earlier and harder than it hits their loan coupons.

Blackstone's counter-argument deserves airtime: on the July 23 call, president Jon Gray said redemption requests at BCRED slowed "materially" in July and attributed the earlier surge to a diminishing "level of noise" in the media (reporting on the Q2 call; see also Private Debt Investor). My opinion: that is probably true and largely irrelevant. Requests slow after a gate for a mechanical reason — investors who know they will be prorated to 5% stop queueing and start looking for a bid. The demand does not disappear; it migrates to the secondary market. Which is exactly what the Cox Capital tender and the Ares stake sale are.

3. The 0.70x problem: one asset class, three prices

Using the BDC price/NAV data in Raymond James' BDC Weekly Insight (data as of July 23, 2026), the listed sector's median price-to-book is roughly 0.70x. Two-thirds of the 44 listed names with clean data trade below 0.75x book. Only six trade above 1.00x.

Jul 23 price and P/NAV: Raymond James BDC Weekly Insight. Jul 28 prices: intraday quotes timestamped 16:00 UTC.

Now line up the three prices this asset class simultaneously carries:

4. Primary source read: Blue Owl Capital Corporation's own filing

Deep-dive standard requires reading the document, not the summary. I read the Blue Owl Capital Corporation (OBDC) Form 40-33 filing. Four things stand out, and they are in the company's own risk language, not a short seller's deck:

• The valuation allegation is now inside the company's own disclosure. The filing records that in March 2026 a Los Angeles-based investment fund alleged Blue Owl's private credit funds misrepresented loan loss rates — specifically that values assigned to loans in OBDC's portfolio were inflated relative to the current public trading prices of those same credits. That is the entire private credit bear thesis, restated in a company filing.
• Reported NAV of $14.81 per share against a market price of $11.05 as of midday today — 0.75x book. The filing's share repurchase reference price of $11.70 sits above where the stock actually trades.
• Non-accruals of $181 million disclosed on the loan book.
• Overconcentration in software and technology is named as a portfolio risk factor. This matters far more in July 2026 than it did in 2024: software and technology borrowers are the borrowers whose end-demand is most levered to the AI capex cycle that the equity market spent today repricing violently (QQQ -0.51%, chip complex extending a multi-day rout per CNBC).

My opinion: the software concentration is the most underpriced line in the entire BDC complex. Direct lenders financed a decade of recurring-revenue software LBOs at 6–8x EBITDA on adjusted numbers. If AI compresses seat-based software pricing — the thesis the equity market is now trading aggressively — those loans do not default next quarter. They stop growing into their leverage, then they get amended, then they get marked. The lag is 4–6 quarters. That is the tail risk that neither the 0.6% index non-accrual rate nor the 6.0% Fitch default rate captures.

5. The measurement scandal: 0.6%, 2.4%, 5.2%

Ask three sources what fraction of private credit loans are non-performing and you get three answers that differ by an order of magnitude:

Raymond James' own table shows a long list of BDCs reporting zero non-accruals as of late July. My opinion, stated plainly: a market where reported credit impairment ranges from 0.0% to 6.0% depending on the counting convention does not have a credit problem yet — it has a measurement problem now, which becomes a credit problem the moment a transaction forces one convention to win. The Cox Capital tender and the Ares stake sale are how a convention gets chosen. The single most important consequence of Fitch's 0.9x dividend coverage figure: at coverage below 1.0x the sector is funding distributions out of capital, which mathematically compresses NAV every quarter it persists, independent of credit losses. That is why 15–20% headline yields at 0.5x book are not value — they are a liquidation schedule with a dividend attached.

6. Second- and third-order effects most people are missing

(a) The GP that sells stakes in its own fund creates a public price for its own marks

This is the un-traded consequence of the Ares deal. If $3.4B of Ares Capital Europe IV changes hands at, say, 92 cents, then every European direct lending fund of that vintage has a comparable. Private marks do not get dragged to market by defaults — they get dragged there by transactions. Ares is a sophisticated seller doing the right thing for its LPs, and in doing so it is handing the entire industry a benchmark it did not ask for.

(b) Fund-finance is the transmission channel into banks

Dechert's July 2026 CFO market update spells out the plumbing: collateralized fund obligations and NAV loans run loan-to-value tests measured against the NAV of the underlying portfolio, and an LP that cannot meet a capital call faces "loss of distribution rights and forced sale of LP interest at a discount to NAV." CLO equity is increasingly common collateral. Put those two facts together with a repricing secondaries market and you get reflexivity: lower secondary marks → LTV breaches → forced sales of LP interests → lower secondary marks. The Financial Stability Board's May 2026 report on private credit vulnerabilities flags exactly this bank-to-fund exposure channel. The equity market is not pricing it at all: XLF made a new high today at 57.53 intraday, +1.00% to 57.45.

(c) The retail distribution build-out is a cycle top signal, not a growth signal

Apollo and Ares began offering European retail clients access to private credit and infrastructure funds through Revolut for as little as €1 — announced seven weeks after the largest retail private credit vehicle in the world capped withdrawals at 5% of shares. Building a wider on-ramp while the off-ramp is metered is a distribution-cycle decision, not a credit decision. Historically, when an illiquid asset class reaches the €1 minimum, the returns have already been earned by someone else.

(d) 53 unprofitable BDCs is a small-business signal, not just a credit signal

Listed BDCs are the marginal lender to mid-market America. If the entire cohort lost money in Q1 2026 for the first time on record — an average loss of $7.6M against a $26M profit a year earlier — then the cost and availability of mid-market credit is tightening at exactly the moment the Fed is being pushed hawkish by energy and tariff pass-through (CNBC on Warsh's bind). That is a Russell-2000 earnings problem in Q4 2026 and H1 2027, not a headline default problem now. IWM at 293.39 (+0.16%) is not discounting it.

7. Today's tape — and why the BDC bounce is a rotation, not a repair

Note the shape: the lowest-quality BDCs led (TCPC +3.80%, HTGC +2.03%, MAIN +1.99%, GSBD +1.95%) while high-yield credit itself was flat and AAA CLO paper literally did not move. When the junk end of an asset class outperforms while the senior end of the same asset class is unchanged, that is beta-chasing out of a collapsing sector (chips) into a cheap one — not credit improvement. The BDC ETF is up 1.44% today and down ~23% year to date. Only one of those two numbers is information.

8. Three scenarios into year-end

9. Positioning — specific levels

1. LONG ARCC HIGH CONVICTION — reiterating, and the referee prints tomorrow

• Price now: $19.04 (+0.79%). Entry zone: $18.40–19.20 (unchanged from July 27 — currently live).
• Target: $21.50 (~1.05x forward NAV). Timeframe: 1–3 months.
• Invalidation (one measurable condition): a daily close below $17.90, or a Q2 report showing NAV down more than 1.5% QoQ with non-accruals above 2% at cost.
• Why two independent signals agree: (a) ARCC is the largest permanent-capital originator and the direct beneficiary of gated competitors withdrawing origination capacity; (b) at 0.95x book it is the only large BDC whose valuation does not require a NAV markdown to be justified — the sector median is 0.70x, so the relative premium is already the market paying for balance-sheet quality.
• Event: Q2 results and call Wednesday July 29, 12:00 ET — two hours before the FOMC decision. Consensus is roughly $0.47 EPS, down ~6% YoY. Do not add before it. This print is the referee for the entire complex.

2. SHORT FSK HIGH CONVICTION — the cheapest thing that deserves to be cheap

• Price now: $10.97 (+1.11%). Entry zone (short): $10.80–11.25.
• Target: $9.20 (~0.47x book — TCPC's current multiple). Timeframe: 1–3 months.
• Invalidation: two consecutive daily closes above $11.85; or a Q2 print (Aug 6) with NAV down less than 2% QoQ and flat non-accruals.
• Why two independent signals agree: (a) its own income statement is already broken — trailing EPS of -$1.96, revenue -15.7% YoY, three-year EPS growth -50.6%, forward P/E of 275x on a 15.9% stated yield; (b) sector dividend coverage of 0.9x (Fitch 1Q26) means the payout is being funded from capital, which compresses NAV mechanically every quarter regardless of new credit losses. A 15.9% yield at 0.56x book is not value; it is the market correctly pricing a return of capital as a return.
• Quantified risk: the squeeze risk is real and today proved it — the junk BDC cohort rallied 2–4% on rotation flows with zero fundamental news. Size accordingly and respect the $11.85 stop.

3. SHORT TCPC SPECULATIVE — reiterate, currently against me

• Price now: $3.28 (+3.80% today). Entry zone: $3.25–3.55 (raised from $3.15–3.45 given today's move). Target: $2.40. Invalidation: daily close above $3.75. Timeframe: 1–3 months.
• 0.47x book, a 21.7% stated dividend yield, a CEO departure at the affiliated BlackRock private credit fund during a valuation probe, and a Q2 print on Aug 6. The thesis is that a 21.7% yield is a countdown, not an income stream. Marked SPECULATIVE, not HIGH CONVICTION, because it is a single-signal short (valuation/quality) and it has already moved against me 3.8% in one session.

4. THE PAIR HIGH CONVICTION — this is the highest-quality expression of the thesis

• Long basket (equal weight): ARCC $19.04, BXSL $23.22, MAIN $54.87. Short basket (equal weight): FSK $10.97, TCPC $3.28.
• Target: 12 points of relative outperformance. Timeframe: 1–3 months. Invalidation: 6% adverse relative move from initiation, or a broad sector re-rating above 0.85x median book (the bull case).
• Why: it isolates the exact variable the secondaries prints are repricing — the ability to fund and originate through a liquidity squeeze — while neutralizing the rate and beta exposure that made the outright shorts painful today. Note MAIN at 1.59x book is the expensive leg and the reason this is a pair, not a naked long.

5. OBDC WATCH — do not buy the 11.4% yield before Aug 5

• $11.05 against a filed NAV of $14.81 = 0.75x. Accumulation zone only after the Aug 5 print: $10.20–10.80. Target: $12.60. Invalidation before entry: reported NAV below $14.00, or non-accruals above $220M (vs the $181M in the filing).
• The trigger I need is NAV stability, not yield. The filing's own acknowledgement of the March 2026 loan-loss-rate allegation plus named software/tech overconcentration means the next two marks carry genuine headline risk.

6. ARES WATCH — the seller is also the toll booth

• $128.65 (+0.70%). Buy zone: $116–124 on a Fed-shock or a post-earnings flush. Target: $148. Invalidation: daily close below $112.
• The contrarian observation: Ares is simultaneously the largest seller of private credit LP stakes this month and the operator of one of the biggest dedicated credit-secondaries platforms, in a market its own research projects growing from $15B (2024) to $28B (2026) to $50B+ (2030). If the repricing I describe happens, Ares collects fees on both sides of it. Offsetting: Morgan Stanley is at Hold, the multiple is full, and reports this week. Hence WATCH, not a long.

7. AVOID / short-on-strength: OWL above $7.75 WATCH

Retail fundraising has evaporated, OBDC sits at 0.74–0.75x book, and the March 2026 loan-loss allegation is now referenced in its own BDC's filing. The prior session's "avoid above $7.75" stands unchanged; I am not asserting a fresh intraday level I have not verified today.

10. Scorecard — honest marks on the open book

The July 24 long/short pair remains net negative. I said so last session and it is still true. The correction I am making today is directional conviction on the short leg: I under-committed to FSK last week on valuation-support grounds, and the income statement says that support does not exist.

11. What would change my mind

• An Ares Capital Europe IV stake sale printing at 96%+ of NAV. That would be strong evidence the marks are close to right and the listed discount is a liquidity artifact, not a valuation signal. I would cut the shorts immediately.
• ARCC reporting NAV up QoQ with non-accruals below 1.5% at cost tomorrow. Best originator, cleanest book — if it is genuinely clean, my software-concentration timeline is too aggressive.
• Q3 redemption requests below the 5% cap at all three of BCRED, Apollo Debt Solutions and ASIF. That kills the rolling-queue mechanism, which is the load-bearing element of the whole thesis.

Summary — the thesis in five lines

1. Private credit's problem migrated from the loan book to the exit door, and in the last five weeks the exit door printed a price: 15–30% below NAV for retail BDC shares, ~90–95% of NAV for institutional stakes.
2. $15.6B of Q2 redemptions were requested and only 38% honored — roughly $9.7B of unmet demand rolling into Q3, which is what forces price discovery into the secondaries market.
3. Every point of NAV haircut across ~$1.8T is $18B; transactions already completed imply $90–180B of unrecognized markdown, before any incremental default.
4. The listed sector's ~0.70x median book is the honest number. Two-thirds of BDCs trade below 0.75x while the non-traded complex still reports 100 cents.
5. Actionable: long ARCC / BXSL / MAIN against short FSK / TCPC; do not add anything before ARCC's 12:00 ET print on July 29 and the FOMC decision two hours later.