Explainer · Private credit

Semi-liquid funds, explained

How interval funds, non-traded BDCs, ELTIFs and evergreen funds let you out — and when they don't

Written 3 September 2026 · All explainers

How to read the markers. [D] — verified from a source retrieved for this note, listed under Sources. [I] — industry-understood; not formally disclosed by a named party. [S] — our own inference or judgment. Every figure is either sourced or marked as an estimate.

The short version

A “semi-liquid” fund holds assets that cannot be sold on demand — private loans, toll roads, data centres — and offers investors a way out anyway. The way out is never the asset. It is the fund’s own plumbing: cash the loans throw off, new subscriptions, a credit line and, last, selling loans to another buyer. Every wrapper is a set of rules about how much of that plumbing investors may drain, and how often.

The rule that matters is the cap: the largest share of the fund the manager will buy back per window, and what happens to requests above it. In a US interval fund the cap is a legal band of 5% to 25% of outstanding shares per offer, set by Rule 23c-3 under the Investment Company Act of 1940. In a non-traded business development company (BDC) the 5% quarterly cap is the fund’s own policy, which its board can raise or cut. In a European ELTIF 2.0 the cap is a percentage of the fund’s liquid pocket, read off a table in a Commission regulation. In every case, when more people want out than the cap allows, each investor is paid a pro-rata slice at net asset value (NAV) and the rest waits.

That waiting is a queue, not a default. In the first quarter of 2026 the twelve largest US non-traded BDCs received requests averaging 12.1% of shares against 5% caps and paid 53% of the money asked for, all at NAV, while listed BDCs — same loans, different exit — traded at a median 26% discount to NAV. [D — see Sources] The wrapper decides whether you leave at the fund’s mark or at the market’s.


1. Why the loan cannot be sold on demand

A private loan is a contract between one lender and one borrower. No exchange quotes it, and a buyer must read the credit file first. A stake in a wind farm changes hands after months of due diligence. Selling either quickly means selling to the few buyers who are ready, at whatever price they offer that week. [I]

So a fund that promises liquidity must find the cash elsewhere. Four sources, in the order managers prefer:

  1. Cash the portfolio throws off. A typical private loan refinances in about three years, so roughly a third of a lending book turns into cash each year. [D — iCapital]
  2. New subscriptions. Cheapest and least reliable: inflows fall in exactly the quarters when requests rise.
  3. Borrowing. Blue Owl Credit Income Corp reported $11.3bn of liquidity across cash, undrawn debt and liquid assets at the end of February 2026. [D]
  4. Selling assets. The last resort, because it fixes a price. In February 2026 Blue Owl BDCs sold $1.4bn of loans to institutions at 99.7% of par. [D]

The cap exists to protect those four sources. A fund that let everyone out at once would sell loans into a thin market, and the investors who stayed would eat the discount. The cap protects the stayers, which is why every wrapper below has one. [S]


2. The US interval fund, and its looser cousin

An interval fund is a closed-end fund registered under the 1940 Act that adopts, as a fundamental policy, a promise to buy back shares at set intervals. Rule 23c-3 sets the terms, and they are law, not house style: [D — Rule 23c-3; SEC investor bulletin, 25 September 2020]

  • Interval: every 3, 6 or 12 months.
  • Offer size: at least 5% and at most 25% of outstanding shares per offer.
  • Timing: notice 21 to 42 days before the deadline; price is the NAV struck within 14 days after it; payment within 7 days of pricing.
  • Oversubscription: the fund may buy up to an extra 2% of shares, then must reduce every request pro rata.
  • Cover: liquid assets equal to at least 100% of the offer amount; repurchase fee of at most 2%.
  • Suspension: only in narrow cases (loss of tax status, market closure, inability to value) or with SEC permission.

The Cliffwater Corporate Lending Fund’s offer of 5 February 2026 shows the rule at work: 5% of shares, discretion to add 2%, pro rata above 7%, NAV on 10 March, payment within seven days, no fee. [D] Investors tendered 13.9% of shares; the fund bought 7% and pro-rated the rest. [D — PitchBook, 12 March 2026] That is the interval-fund promise: it can be pro-rated but it cannot be gated — the offer must happen, at the stated size, on the stated date. [D — XA Investments, Q1 2026]

The tender-offer fund is the looser cousin. Also a closed-end fund, it buys shares back through issuer tender offers under Rule 13e-4 of the Securities Exchange Act of 1934, whenever its board chooses. Rule 13e-4 says how a tender must run — open at least 20 business days, pro rata if oversubscribed — but sets no minimum frequency and no minimum or maximum size. [D — Rule 13e-4; Chapman] Most offer 5% a quarter; in law they need not offer anything. They are also more often sold privately to accredited investors. [D — Chapman]

At 31 March 2026 XA Investments counted 173 interval funds and 141 tender-offer funds, with $247bn of net assets between them. [D]


3. Non-traded BDC versus listed BDC

A BDC is a closed-end fund that elects a special status under the 1940 Act, built for lending to private US companies. [I] The same strategy comes in two exits.

A listed BDC sells its shares once; they then trade on an exchange. The fund never buys them back on demand. You exit by finding a buyer, at whatever price the market sets — above NAV (a premium) or below it (a discount). [D — SEC investor bulletin on closed-end funds; ICI]

A non-traded BDC never lists. It sells shares continuously at NAV and buys them back through quarterly Rule 13e-4 tenders. The cap is the fund’s own policy. Blackstone Private Credit Fund (BCRED), the largest, puts it this way in its 2024 annual report: the fund “may repurchase, in each quarter, up to 5% of the NAV of the Company’s Common Shares outstanding”, at NAV, less a 2% early-repurchase deduction on shares held under a year; and the board “may amend or suspend the share repurchase program at any time”. [D — BCRED 10-K for 2024] Buyers need $250,000 of net worth, or $70,000 of income plus $70,000 of net worth; the minimum ticket is $2,500. [D — BCRED fact sheet, 31 December 2025]

Two things called “5% a quarter” are not the same thing. The interval fund’s 5–25% band is law, per offer, as a share of outstanding shares. The non-traded BDC’s 5% is a board promise: BCRED raised it to 7% in early 2026 and could equally have cut it. [D] Only one is enforceable by a regulator.

Why the listed price sits below NAV while the non-traded fund pays NAV

NAV is the manager’s valuation of loans nobody is trading. The exchange price is what a stranger will pay today for a claim on those loans. When they diverge, the market is disagreeing with the mark. At end-March 2026 the median listed private-credit BDC traded at 0.74 times forward NAV — a 26% discount, the widest since October 2020. [D — LSEG via Reuters, 23 April 2026] PIMCO’s reading in August 2026: listed discounts “reflect scepticism toward reported marks”. [D — PIMCO, 17 August 2026]

Non-traded BDCs kept paying 100% of NAV (98% for early leavers) throughout. So what? The discount is a public estimate of what the private marks may be missing; an investor leaving at NAV is paid a price the listed market will not pay. That is itself a reason to queue, which pushes requests above the cap. The wrapper that pays the higher price is the one that has to ration. [S]


4. Europe’s ELTIF 2.0

Regulation (EU) 2023/606 rebuilt the European Long-Term Investment Fund and applies from 10 January 2024. It removed the €10,000 minimum ticket and the 10% portfolio limit for retail investors, replacing them with a MiFID II suitability assessment. [D — EUR-Lex] The redemption rules sit in Article 18 and in Commission Delegated Regulation (EU) 2024/2759, the regulatory technical standards (RTS), published on 25 October 2024 and in force from 26 October 2024. [D — EUR-Lex; Norton Rose Fulbright]

Article 18 starts from a default of no redemptions before the fund’s end date, then allows them on five conditions: any minimum holding period has passed; the manager has a redemption policy and liquidity-management tools fit for a long-term strategy; the policy states the procedures; redemptions are limited to a percentage of liquid assets; and excess requests are met pro rata. [D — Reg. 2023/606, Art. 18(2)] The RTS then fixes the numbers:

  • Minimum holding period: not required and no length set; if chosen, Article 3 says it should normally cover the fund’s initial investment phase. [D — RTS Art. 3, recital 4]
  • Frequency: any, but more often than quarterly must be justified to the national regulator. [D — RTS Art. 5(4)]
  • Notice period: any, but under three months must be explained to the regulator. [D — RTS Art. 5(8)]
  • The cap: a maximum percentage of the fund’s liquid assets plus its prudently forecast 12-month cash flow — explicitly excluding hoped-for new subscriptions — read from Annex I (by frequency and notice period) or Annex II (by frequency and minimum liquid holding). [D — RTS Art. 5(5)–(6)]
  • Anti-dilution tools: the manager “shall not be required but may” adopt levies, swing pricing or redemption fees. ESMA’s December 2023 draft made one mandatory; the adopted text did not. [D — RTS Art. 5(9); ESMA, 19 December 2023]
Redemption frequency Minimum liquid assets (Annex II) Max redemption, % of liquid pool (Annex II) Same frequency, 3-month notice (Annex I, option 1) Source
Annual or less often 10% 100% 100% Delegated Regulation 2024/2759, Annexes I–II
Every 6 months 15% 67% 66.7% same
Quarterly 20% 50% 33.3% same
Monthly or more often 25% 20% (monthly aggregate) 11.1% (monthly) same

So what? The ELTIF cap is measured against the liquid pocket, not the whole fund. A quarterly ELTIF holding the Annex II minimum of 20% in liquid assets can pay out at most half of that pocket — 10% of the fund — per quarter, and cannot count new inflows toward it. That is stricter by design than a US non-traded BDC, which may lean on subscriptions. [S — arithmetic on the RTS tables] The wrapper is small but growing: at least €34.0bn across 268 registered ELTIFs at end-2025, up 55% on the year, with 113 launched in 2025. [D — Scope, March 2026]

Four semi-liquid wrappers compared: interval fund, tender-offer fund, non-traded BDC, ELTIF 2.0 A table with four columns and five rows: legal basis, who can buy, how often the window opens, what caps the window, and the exit price. All four wrappers pay net asset value on exit and none is listed. A contrast strip at the bottom shows the listed BDC, which pays the exchange price — a median 26% below NAV at end-March 2026. Four wrappers · one question: how do you get out? Interval fund US · 1940 Act Tender-offer fund US · 1940 Act Non-traded BDC US · 1940 Act BDC election ELTIF 2.0 EU · Reg. 2015/760, amended Cap set by Legal rule · Rule 23c-3 fundamental policy Board discretion · Rule 13e-4 no legal min or max Fund's own policy · Rule 13e-4 board may amend or suspend Regulation + RTS 2024/2759 manager picks Annex I / II Who can buy Any investor registered, publicly offered Often accredited only many are privately offered Retail with suitability test e.g. $250k net worth · $2.5k retail after MiFID test no €10k minimum from 2024 Window opens Every 3, 6 or 12 months must happen, no skipping When the board decides quarterly is common practice Quarterly, at board discretion may be reduced or suspended Set by the fund monthly needs justification Cap per window 5–25% of shares outstanding +2% discretion · then pro rata Whatever the offer says typically 5% · pro rata above it Typically 5% a quarter BCRED: raised to 7% in Q1 2026 share of liquid pocket quarterly, 3m notice: 33.3% Price you exit at NAV fee at most 2% NAV terms per offer NAV 98% if held under one year NAV anti-dilution tools optional none of the four is listed Listed BDC, for contrast Market price on an exchange · the fund never buys back on demand median listed private-credit BDC at end-March 2026: 0.74 × NAV, a 26% discount (LSEG via Reuters)
Figure 1. Four semi-liquid wrappers compared on who can buy, how often the window opens, what caps it, and the price you leave at. All four pay net asset value; the listed BDC, for contrast, pays whatever the exchange offers. Caps from Rule 23c-3, Rule 13e-4, the BCRED 10-K and Delegated Regulation 2024/2759; the 26% discount is the LSEG median for listed private-credit BDCs at end-March 2026.

5. The gate, worked through

A gate is not a refusal. It is a queue with a fixed service rate. Take a fund of 1,000 shares at $1m each — $1bn — with a 5% quarterly cap, in a quarter when holders of 120 shares (12%) ask out. Assume NAV stays flat and nobody new joins the queue. All numbers are illustrative.

Quarter Shares outstanding at start Tendered Cap (5%) Paid Each request filled Cohort out, cumulative
Q1 1,000 120 50 50 41.7% 41.7%
Q2 950 70 47.5 47.5 67.9% 81.3%
Q3 902.5 22.5 45.1 22.5 100% 100%

Three points the arithmetic makes:

  1. Everyone is paid something; nobody is paid everything. In Q1 each $1m request receives $417,000 at NAV. That is what “pro rata” means.
  2. The remainder does not queue automatically. Ares Strategic Income Fund’s tender letter says it plainly: “unfulfilled repurchase requests do not carry over automatically to the next repurchase offer.” [D] The investor must re-tender, alongside whoever else has arrived.
  3. The cap shrinks with the fund. 5% of 950 is less than 5% of 1,000; with no inflows, each quarter pays a little less.

If a fresh 12% asks out every quarter, the queue never clears: requests arrive at 12% and leave at 5%, so the backlog grows by about 7% of the fund a quarter and the fill rate falls. Several BDCs were in that regime through 2026. [S — illustrative]

The gate as a queue: a 5% quarterly cap against a 12% request, over three quarters Three bars, one per quarter. In the first quarter holders of 120 of 1,000 shares tender; the fund pays 50 (the 5% cap) and 70 roll forward to be re-tendered. In the second quarter 70 are tendered against a cap of 47.5; 22.5 roll forward. In the third quarter 22.5 are tendered against a cap of 45 and all are paid. Each request is filled 41.7%, then 67.9%, then 100%; the cohort is fully out after three quarters. Illustrative numbers. The gate as a queue · illustrative $1bn fund · 5% quarterly cap · 12% of shares tendered in Q1 · nobody new joins shares (of 1,000 at start) Q1 · 120 tendered 70 unfilled must re-tender 50 paid at NAV cap 50 41.7% of each request filled cohort out: 41.7% 70 roll forward Q2 · 70 tendered 22.5 unfilled 47.5 paid cap 47.5 67.9% of each request filled cohort out: 81.3% 22.5 roll forward Q3 · 22.5 tendered 22.5 paid cap 45 100% of each request filled cohort out: 100% paid this quarter, at NAV unfilled · rolls to next offer cap Everyone is paid something · nobody is paid everything · the cap shrinks with the fund
Figure 2. The gate as a queue. A $1bn fund with a 5% quarterly cap receives requests for 12% of its shares; each quarter it pays the cap and the rest must be re-tendered. Illustrative round numbers, flat NAV, no new requests after Q1. On those assumptions the first-quarter cohort is fully out after three quarters.

6. What 2025–2026 showed

Pressure built through late 2025: across the non-traded BDCs in Cliffwater’s index, redemptions rose from 1.6% of NAV in the third quarter to 4.8% in the fourth. [D — CAIA, citing Cliffwater] The first quarter of 2026 broke through the cap.

Fund (sponsor) Quarter Requested, % of shares Cap applied Paid Source
Blue Owl Technology Income Corp Q1 2026 40.7% 5% pro rata Investment Executive; CAIA
Blue Owl Credit Income Corp Q1 2026 21.9% 5% ~23% of each request; $988m Blue Owl 8-K, 2 Apr 2026
Cliffwater Corporate Lending Fund (interval fund) Q1 2026 13.9% 7% (5% + 2%) pro rata Cliffwater N-23C3A; PitchBook
Ares Strategic Income Fund Q1 2026 11.6% 5% 43.1%; $524.5m Ares tender exhibits, Mar 2026
Apollo Debt Solutions BDC Q1 2026 11.2% 5% ~45%; ~$730m With Intelligence; Investment Executive
Oaktree Strategic Credit Fund Q1 2026 8.5% raised to 7%; affiliate bought 1.7% 100% Reuters, 27 Mar 2026
BCRED Q1 2026 7.9% raised to 7%; Blackstone put in $400m 100%; $3.7bn PitchBook, 3 Mar 2026; BCRED letter, 4 Jun 2026
BCRED Q2 2026 ~10% 5% pro rata BCRED letter, 4 Jun 2026
BCRED Q3 2026 ~10%; $4.3bn 5% pro rata Reuters, 3 Sep 2026

Across the twelve largest non-traded BDCs, first-quarter requests averaged 12.1% of shares (median 10.1%), about $15bn in total, of which 53.4% was paid. [D — With Intelligence, 30 April 2026] Stanger’s count: $4.9bn raised, $6.9bn of redemptions paid, and “for the first time, more capital left non-listed BDCs in a quarter than came in”. [D — Stanger, 14 May 2026] Second-quarter payouts were $5.9bn, taking the half-year to $12.7bn, while April BDC sales of $1.6bn were down 74% on the year. [D — Stanger, 6 July and 26 May 2026]

Three mechanisms showed themselves that had only been on paper before:

  • The board’s cap is elastic both ways. Blackstone lifted BCRED’s cap to 7% in Q1 — “the maximum amount permitted without changing the terms of the repurchase offer” — then held 5% in Q2 and Q3 against requests of about 10%. Apollo never moved off 5%. [D]
  • Sponsors can shorten the queue from outside. Blackstone and its employees put $400m into BCRED; Brookfield, Oaktree’s parent, bought 1.7% of Oaktree Strategic Credit Fund’s shares from a tendering investor so every request was met. [D] Neither is an obligation; both were choices.
  • The wrapper can be rewritten. Blue Owl’s OBDC II board replaced future tender offers with quarterly return-of-capital distributions to every holder — no queue at all. [D — Blue Owl 8-K, 18 February 2026]

So what? The promise was kept to the letter: every offer ran, every accepted share was paid at NAV, no fund suspended repurchases. What investors discovered was the size of the letter. Stanger’s chairman gave the manager’s view: “The semi-liquid structure allows them to do so on defined terms, protecting the investors who want to stay without forcing distressed sales inside the fund.” [D — Stanger, 6 July 2026] The investor’s view is section 5: at 5% a quarter against 10% of requests, half of each request waits, and the wait compounds.


Caveats

  • Sources disagree on how many BDCs pro-rated in Q1 2026. With Intelligence says seven of the twelve largest; Stanger says five. The universes differ and neither publishes its list. Both are recorded; neither is smoothed.
  • BCRED’s size is quoted inconsistently. Reuters and PitchBook give $77–82bn; BCRED’s own fact sheet gives net assets of $47.6bn at 31 December 2025. The larger figures appear to include borrowing.
  • The Oaktree figures are as reported by Reuters, reconciling the fund’s 6.8% repurchase and the affiliate’s 1.7% purchase to the 8.5% requested. The fund’s own filing for the quarter was not located.
  • Tender-offer fund eligibility is practice, not rule. Chapman says they are “often” privately offered to accredited investors; some are public.
  • Whether ELTIFs list is treated as “usually not” [I]. The regulation permits listing; the share that do was not retrieved.
  • The section 5 example assumes flat NAV, no inflows and a single cohort. It shows the queue mechanism, not any fund’s outcome.
  • The ELTIF market size is a floor. Scope reports “at least” €34.0bn because 30 of the 268 registered funds disclose no volume.
  • The claim that widening listed discounts push NAV-exit requests above the cap is [S], our inference from the two price series.
  • All fund figures are point-in-time to 3 September 2026 and will be superseded by the next quarter’s tender results.

Sources [D]

All retrieved and confirmed to load on 3 September 2026.

US rules and regulator guidance

EU rules

Fund documents (SEC EDGAR unless stated)

Market data and trade press