Private Credit · Standalone Deep Dive

The $3.5 Trillion Market That Has Started Buying Its Own Loans — and Setting Its Own Price

Only about 1% of private credit NAV gets price-tested by an outside buyer each year. That 1% is increasingly bought by the same firms that sold it. Two documents published this summer are about to make that impossible — and that is the catalyst nobody is trading.

Friday, 21 August 2026 · @dailyanalysts · Prices as of the 21 Aug US session

1. The event: a lender raising a fund to buy its own loans

On 19 August, Jefferies Credit Partners went to market for roughly €1bn — about $1.16bn — for a private credit secondaries fund. The strategy has two legs: write new loans, and buy loans out of Jefferies' own existing credit portfolios. Jefferies Credit Partners already runs about $25bn across direct lending and CLOs, and sits inside Jefferies Finance, the 50/50 JV with MassMutual (Benzinga, Crypto Briefing).

The pitch is candid about why this is attractive: because the fund can source from Jefferies' own book, it "has a built-in source of deal flow that doesn't depend on winning competitive auctions against other secondary buyers." Read that sentence again. The selling point is the absence of a competitive auction.

It is not an outlier. In the last six months the same structure has been printed repeatedly:

The single most important number in this piece. Carlyle/AlpInvest estimate that only ~$20bn of credit fund exposures traded in 2025 against a direct lending market measured in the trillions — a turnover rate of roughly 1%, versus 2–3% in private equity secondaries (Carlyle/AlpInvest, June 2026; summarised by The Credit Crunch).

The consequence: roughly 99% of private credit NAV has never been shown to an arm's-length buyer. The 1% that has is increasingly transacted between affiliates. That is the entire basis on which $3.5trn of marks — and every management fee computed off them — currently stands.

2. What the 1% actually clears at (and why the headline is a lie)

The industry markets credit secondaries pricing as strong and getting stronger: 91% of NAV at end-2024, 92% in H1 2025, per Jefferies' own secondary market review. Senior direct lending interests are commonly quoted at 95–100% of NAV. Directionally that supports the marks.

Then read what the people executing these deals told PitchBook LCD about how the arithmetic actually works on the half-dozen $1bn-plus credit continuation vehicles done so far:

A deal that prices nominally above par — for example, 102 — may get three points of discount from the portfolio's NAV (which may be marked at 97) and another four points from the two quarters of post-reference-date cash flows that have accrued to the buyer's benefit. This brings the indirect discount to seven points, and an effective price of 95, on a deal with a headline price of 102.

And the kicker, from the same reporting: no deal has traded at a true premium to par. One banker put the highest effective price ever seen in the market at par, at most.

So here is the real clearing price of private credit, triangulated three ways: Call it 92–95 cents. Every mark in the industry is carried at approximately 100. The book is roughly five points rich — not thirty, as the loudest bears claim, and not zero, as the managers claim.

Five points does not sound like a crisis. It is not meant to. But apply it: five points on the ~$561bn of BDC assets is ~$28bn of net asset value; five points on the $3.5trn complex is ~$175bn. And it lands almost entirely outside the regulated banking system — on LP pensions and endowments, on PE-owned insurance balance sheets, and on the retail investors in non-traded BDCs.

3. Why the self-dealt mark is about to get harder to print

Two documents published within five weeks of each other are, in my view, the most consequential development in private credit this quarter — and neither has moved a single price.

(a) ILPA's June 2026 draft Continuation Vehicle Guidance

ILPA's draft guidance — dissected well by Mercer Capital's Jeff Davis — moves the standard from validating a price to policing a process. CVs were ~14% of sponsor-backed exits globally in 2025, up from 5% in 2021. The prescriptions:

(b) CFA Institute, 23 July 2026: "Conflicts of Interest in Continuation Funds"

Stephen Deane's report opens with the sentence the industry has spent three years not saying out loud:

In efficient capital markets, prices are determined by arms-length transactions between buyers and sellers. But what happens when the buyer is the seller?

Its core finding is the one that matters for marks: because the GP solicits the bids, selects the winner and negotiates the price, "price discovery in CV transactions is neither objective nor fully independent" — and therefore fairness cannot be judged on price at all, only on process. The report separates structural conflicts (unavoidable), procedural failures (fixable), and intentional misconduct (concealment, manipulation of valuations, selective disclosure).

My read — and this is the non-consensus part. Everyone is treating ILPA and CFA Institute as governance housekeeping. They are a valuation mechanism. The specific requirement to publish an anonymised summary of final-round bids and reconcile the transaction price to reported NAV does one thing: it forces a second, losing, arm's-length bid into daylight for the first time. Lead bids in credit CVs cluster at 92–95. Cover bids do not. The first time an LPAC sees a final-round bid range with an 80-handle in it, the reference price for an entire vintage changes — and it changes in a document the manager is required to circulate.

The reform designed to protect LPs is the mechanism that marks the book down. That is the trade.

4. Why the pressure valve is jamming shut at the same time

The secondaries bid is not just a price-discovery problem; it is a capacity problem, and the supply side is accelerating faster than the demand side.

PressureDataConsequence
Aging vintages need DPI2019-vintage credit funds average ~0.7x DPI (Ares)A few years to convert NAV into cash, or the CV becomes the only exit
Retail wants outQ2'26 non-traded BDC redemption requests averaged 10.3% of shares; 10 of 16 tracked funds capped at the 5% limit; gross inflows −56% QoQ (Fitch). HPS's $26bn HLEND capped 13.3% of requests at 5%. Apollo ADS onshore 4.3% vs offshore 12.5%Gated capital has to be met by selling loans — into the secondaries bid
The bid is tiny~$20bn traded in 2025. Apollo projects $50bn within 2–3 years; Carlyle $80bn+ by 2030 (~$55bn GP-led, ~$25bn LP-led)Even the bull case is ~2% annual turnover on the asset base. This market cannot absorb a genuine wave — it can only price the trickle
The buyer is the sellerJefferies buying Jefferies; BlackRock TCP → Pantheon CV; Ares → Coller CVThe "liquidity" is intramural. Correlation across the complex is higher than any allocator's model assumes

Meanwhile the underlying credit is not improving. Fitch's US private credit default rate hit a record 6.0% TTM in Q2'26 and remained at that record in July. PIK now runs on ~10% of BDC loans, up from ~6% in 2022 — while spreads simultaneously compressed about 1pp to 4–5pp over SOFR, a contradiction the Boston Fed flags as evidence of implicit restructuring or eroding standards. Their tradable finding: a 1-SD decline in a BDC's fair-value ratio predicts ~50bp of negative abnormal equity return the following quarter. Screen the delta, not the level.

5. Where today's prices actually sit

Ticker21 AugChgP/NAVNote
OBDC Blue Owl Capital Corp$11.31+0.27%0.79xNAV $14.26; yield 10.66%; 13 insider buys / 0 sells
FSK FS KKR$11.99+0.67%0.66xNAV $18.30; non-accruals 3.8%; $300m buyback
TCPC BlackRock TCP$4.06+0.74%0.62xSold $523m into Pantheon CV at 95c; DOJ valuation probe
CGBD Carlyle Secured Lending$11.38+0.44%
BXSL Blackstone Secured Lending$24.82+0.61%0.97xCoverage ~97%; NAV down 5 straight quarters; reset telegraphed
ARCC Ares Capital$19.97+0.96%1.03xOnly large BDC above book
BIZD BDC ETF$13.30+0.08%Sector barely participated in the tape rebound
OWL Blue Owl Capital Inc$11.63+2.02%P/E 244x TTM; the fee stream on the same loans
BX / APO / KKR / ARES143.66 / 132.76 / 108.89 / 142.10+1.6 / +2.1 / +1.7 / +1.3%Full reversal of Thursday's 2–3% GP drawdown
SPY / HYG / TLT767.21 / 79.60 / 82.06+0.60% / +0.05% / −0.34%Equities bounce, HY flat, duration still bleeding

Macro overlay (21 Aug): 10Y 4.72%, 30Y 5.26%, both +2bp, with the 30Y still within 7bp of a 19-year high, ahead of Warsh at Jackson Hole 26–28 Aug (CNBC). The consequence for this trade is direct: OBDC's ~9.9% portfolio yield against a 5.26% 30Y is a 4.6pp risk premium — the thinnest of the cycle. 30Y sustainably below 5.00% is the single cleanest catalyst for BDC discounts to close. 30Y above 5.50% invalidates every long in this note on the discount rate alone, regardless of credit.

Note the tape asymmetry, again. On Thursday's −700 Dow day the GPs fell 2–3% while the BDCs holding the actual loans were flat. Today the GPs recovered 1.3–2.1% while BIZD added 0.08%. Fee streams are a high-beta claim on the same collateral the BDCs own at a 20–38% discount. That is the structural inefficiency this entire note is trading.

6. The third-order effect: today's AI paper is tomorrow's secondaries problem

While the industry works out how to price 2019-vintage software loans, it is originating the next problem at ten times the size. Broadcom is in talks with lenders including Blackstone and Apollo for more than $60bn, with subordinated tranches potentially taking the package to ~$100bn, to fund custom AI chips and infrastructure (SiliconAngle, 20 Aug). It builds on the $35bn AI XPV platform Apollo led with Blackstone in June, enabling 20GW of compute through 2028 and Anthropic's 1GW+ expansion. AVGO closed at $368.01, +1.09%.

Three consequences that follow directly from everything above:

  1. Ratings describe the offtaker, not the collateral. What makes a GPU facility investment grade is a take-or-pay contract, not the chips. Where the payer is an AI lab, that is not credit substitution — it is credit deferral.
  2. The residual value problem is the secondaries problem, five years early. GPUs are displaced every 2–3 years and nobody knows what a used one is worth. When these positions eventually need a secondary bid, there will be no comparable trade and no observable clearing price — the exact condition that produced the 92–95c fog in direct lending, applied to an asset with a depreciating physical residual.
  3. Circularity. Several of the platform managers that originate this paper also own the insurers likely to hold it. A downgrade is therefore not just a credit event; it is an RBC/Solvency II capital event that creates forced sellers into a market with almost no secondary liquidity.

7. What I would do — positions, levels, invalidations

High Conviction  LONG OBDC — Blue Owl Capital Corp (reaffirmed)

Why: three independent signals. (1) The clearing price of comparable first-lien collateral, established by arm's-length buyers, is 92–95c — OBDC's equity is priced at 79c on book. (2) Senior/junior disagreement: BDC IG bonds are 1.5–4x oversubscribed (PIMCO took 100% of OBDC's $400m 6.4% 2028 notes) while the equity demands a 21% discount; the debt market and the equity market cannot both be right. (3) Insiders: 13 purchases, zero sales, including CEO Craig Packer at $11.75 and director Logan Nicholson at $11.31 on 2 June — above today's price. The real risk here is rates, not credit: coverage is ~103% ($0.34 adjusted NII vs $0.33 dividend); a 100bp SOFR cut removes roughly $0.05–0.07/share/quarter and forces a dividend cut. The bear case for OBDC is Fed easing, which is the opposite of what most people assume.

Speculative  LONG FSK — FS KKR Capital (new)

Why: FSK is the highest secondaries-optionality per dollar in the sector. It trades at 0.66x book — a 34-point discount against a market where affiliated buyers transact seasoned loan books at 92–95. Management has a $300m buyback authorised and dividend coverage at exactly 100%. Any disclosed portfolio sale, CV or LP-stake trade at even 88c prints an instant 22-point arbitrage against the public price. Why only speculative: TTM EPS is −$1.34, ROE −6.7%, revenue −20.1%, and the entire August sell-side action was price-target raises that left the ratings at Neutral/Hold (JPM $11, Truist $12, RBC $13). You are buying a balance sheet and a catalyst, not an earnings stream.

High Conviction  PAIR: LONG discounted BDCs / SHORT OWL (reaffirmed, and it went against me on day one)

Why: long the assets at 0.79x book, short the fee stream on the same loans at 244x trailing earnings. Book value is written down slowly and in arrears; management-fee revenue is written down the instant net assets shrink. If the ILPA/CFA process reforms force even a 3–5 point downward revision in CV reference prices, the BDCs — already discounting 21–38 points — absorb it and the fee-earning AUM does not.

Honest scorecard: I put this on yesterday. Today OBDC +0.27% against OWL +2.02% — the pair is down ~1.75 points in 24 hours, entirely on the short leg, in a broad risk-on bounce that lifted every alternative manager 1.3–2.1%. That is exactly the risk I flagged when I sized it: OWL has already rallied ~25% and the sell-side is warm on it. Trade the pair, never OWL outright, and do not add to the short above $12.30.

Speculative  LONG TCPC — BlackRock TCP Capital (reaffirmed)

TCPC is the purest expression of this entire thesis: it is the one BDC that has already shown the market its clearing price — 95c on gross fair value, to Pantheon, for $523m of its book — and its equity still trades at 0.62x. Net leverage goes 1.38x → 0.4x → below 0.3x; $157.5m cash plus ~$152m of proceeds; a $50m buyback with only $2.87m used. The board explicitly listed "orderly realization of portfolio assets" as an option — liquidation is the bullish outcome at 0.62x book. Why only speculative: non-accruals 7.4% at cost, revenue −27.4% YoY, ROE −19.08%, and having sold 48% of the debt portfolio, NII falls hard. Assume a dividend cut. You are buying the balance sheet and the DOJ-probe-shaped catalyst, not the income.

Watch  BXSL dividend reset · JEF

BXSL ($24.82): coverage ~97%, NAV down five consecutive quarters, Q2 revenue missed ($320.5m vs $327.8m) on an EPS beat. Management has telegraphed the reset. The cut is the buy signal, not the sell signal — OBDC bottomed after its 16% cut restored 116% coverage. Buy zone on announcement $21.50–23.00. Do not chase above $25.50. No trigger yet.

JEF ($12.7bn cap, 14.2x TTM, ROE 8.5%, 52-week range $35.53–$71.04): the secondaries fund adds fee-earning AUM to a $25bn credit platform — and concentrates precisely the conflict the CFA Institute report describes. I am not short it and I am not long it. I am watching whether the first Jefferies-to-Jefferies trade gets an independent price validation disclosed. If it does not, this is the test case litigators will use.

Tripwires that void the longs

8. Three scenarios into year-end

ScenarioProb.Trigger conditionsOutcome
Bull30%ILPA guidance adopted broadly; the first competitive CV processes print final-round bid ranges in the low-to-mid 90s, validating marks; 30Y back below 5.00% after Jackson Hole; Q3 BDC non-accruals stablePublic discounts close toward 0.90x. OBDC $13.20, FSK $14.50, TCPC $5.00. The pair loses money on the short leg but the longs more than pay for it
Base50%Guidance adopted; credit secondaries volume grows toward ~$28bn for 2026; effective pricing drifts to 88–92 as competitive bidding is enforced; non-traded NAVs quietly trimmed 2–4%; 30Y range-bound 5.10–5.40%Public BDCs grind toward 0.85x. OBDC $12.00–12.50, FSK $13.00. The pair works modestly as GP multiples compress on softer fee-earning AUM growth
Bear20%A disclosed final-round bid range carries an 80-handle, or a large non-traded BDC takes a >5% NAV write, or 30Y breaks 5.50%Public BDCs re-rate down with the private marks, not up against them. OBDC $9.50–10.50; FSK sub-$10. This is the state of the world where the short OWL leg pays for everything — OWL to $9.50 or below

9. The thesis in five sentences

  1. Roughly 99% of private credit's $3.5trn of NAV has never been priced by an arm's-length buyer; the 1% that has is increasingly bought by the seller's own affiliate.
  2. Where genuine outside money has transacted — Pantheon at 95c, LCD's reconstruction of effective mega-CV pricing at ~95 on headline 102s, Jefferies' 92% average — the answer is consistent: the marks are about five points rich, not thirty.
  3. ILPA's June guidance and the CFA Institute's July report will force competitive bidding, independent price validation and disclosure of losing bids — converting an internal negotiation into observable price discovery for the first time.
  4. That is bearish for private marks and for the fee streams computed off them, and bullish for public BDC equity already discounting 21–38 points of the same collateral.
  5. Own the assets at 0.62–0.79x book (OBDC, FSK, TCPC); rent the bear case through short OWL; and let the 30-year Treasury tell you when to size up — under 5.00% you press, over 5.50% you are out.

Action steps

  1. Today: establish or add to OBDC in $10.80–11.50. Half-size FSK in $11.40–12.10.
  2. Set the alert: 30Y through 5.00% = press all longs; 30Y through 5.50% = flatten all longs.
  3. Watch the calendar: Jackson Hole 26–28 Aug (Warsh); Q3 BDC 10-Qs and the Stanger/With Intelligence non-traded flow data in early November; Fitch and Proskauer Q3 default prints in late October.
  4. Watch the documents, not the headlines: the first credit CV that circulates an anonymised final-round bid summary is the single most important private-credit datapoint of the next six months. Read the bid range, not the price.
  5. Do not chase OWL short above $12.30, and do not run the short leg unpaired.
Primary and key sources: CFA Institute, "Conflicts of Interest in Continuation Funds," 23 Jul 2026 · ILPA Draft Continuation Vehicle Guidance, Jun 2026 · Mercer Capital on the ILPA draft · Carlyle/AlpInvest, Credit Secondaries, Jun 2026 · Apollo Academy, Private Credit Secondaries, Jun 2026 · PitchBook LCD on effective CV pricing · The Credit Crunch, 16 Jun 2026 · Ares on credit CVs · Coller/Ares CV, 10 Feb 2026 · Benzinga on Jefferies Credit Partners · Boston Fed CPP 26-6 · Fitch on non-traded BDC flows · FSB, Vulnerabilities in Private Credit, 6 May 2026 · OBDC 10-Q (30 Jun 2026) · TCPC 10-Q (30 Jun 2026) · CNBC, Treasury yields, 21 Aug 2026 · SiliconAngle on Broadcom's AI debt package

Analysis and opinions are my own and are not investment advice. Positions described are analytical calls, not recommendations tailored to any individual. Do your own work.