90-Day Notice vs Repurchase Caps: Model Your Fund Exit

A fund can publish a stable net asset value and still be unsuitable for a payment due in three months. The missing variable is the amount of cash an investor can actually receive by the deadline. A notice period, a dealing date, a repurchase limit, and settlement are separate constraints. Treating them as one promise is a liquidity-planning error.

The immediate catalyst is the UK Financial Conduct Authority's October 8, 2026 consultation, CP26/35. It proposes redemption arrangements for certain funds holding inherently illiquid assets. Comments close on December 11, 2026, and the regulator expects final rules in the first half of 2027. These are proposals, not rules already in force. FCA consultation page, October 8, 2026.

The next question for investment committees is operational: how much of an allocation should count toward a dated cash obligation when access depends on notice, limited repurchases, or both? This analysis provides a reproducible model. All monetary examples are hypothetical; none is a forecast for a named fund.

Separate the mechanisms before comparing products

The FCA paper proposes a minimum 90-day notice period and dealing no more frequently than monthly for in-scope Non-UCITS Retail Schemes with at least 50% of scheme property in inherently illiquid assets. It also recognizes that some assets need longer disposal periods. A 90-day minimum is therefore not proof that every portfolio can reliably produce cash in 90 days. FCA CP26/35, October 8, 2026, chapters 1–3.

US interval funds illustrate a different mechanism. The SEC's investor bulletin explains that periodic offers generally repurchase 5%–25% of outstanding shares, with oversubscribed offers generally allocated pro rata. This is a fund-level limit, not an entitlement for each investor to redeem that percentage of their own holding. SEC investor bulletin, September 25, 2020.

The jurisdictions and legal structures are not interchangeable. The table compares constraints, not equivalent products or investment returns. Product documents, applicable rules, and the actual offer determine an investor's rights.

For a cash forecast, record five milestones: request submission, the relevant cutoff, acceptance or allocation, pricing, and payment. A notice period belongs to that sequence; it is not a substitute for the sequence. The operational test is whether money has settled into an account that can meet the liability.

Translate a fund-level cap into investor cash

Consider a hypothetical fund with 100 million shares, each worth $1, and a repurchase offer for 5% of outstanding shares. Its capacity is five million shares. An investor holds one million shares and tenders the entire position.

Let $g$ be the offer fraction and $q$ the fraction of all outstanding shares tendered. Assuming uniform pro-rata allocation, with no special priority or rounding effects:

Here $\alpha$ is the accepted fraction of each request. If aggregate requests represent 20% of shares, the acceptance fraction is $0.05/0.20=0.25$. The investor receives proceeds for 250,000 shares, not for the full one-million-share request.

The following table is calculated from those assumptions. The share price remains $1, fees are zero, and payment is assumed to occur before the investor's deadline.

The model exposes two common misreadings. A 5% offer is not a 5% loss. Nor does quarterly availability imply that an entire holding becomes cash every quarter. Economic value and access to that value must be modeled separately.

The remaining position is also not necessarily a place in a first-in-first-out queue. Whether an unfilled request persists, expires, or requires resubmission depends on the offer terms. A spreadsheet that automatically carries requests forward can create cash flows that the investor has never actually requested.

Repeated offers do not produce a simple exit date

Suppose the investor resubmits all remaining shares at each eligible offer. For illustration, assume the accepted fraction remains constant at $\alpha$, the NAV is unchanged, and there are no distributions, fees, or new purchases by that investor. The balance after $n$ offers is:

This is a scenario identity, not a prediction that aggregate requests will remain stable. To liquidate at least 90% of the starting position:

Even this stylized calculation changes the planning conversation. Under 25% acceptance, four offers release about 68.4% of the original shares, not 100%. The number of offers is deliberately not expressed as a promised calendar exit date: the first eligible window, dealing frequency, settlement timing, and future offer terms all matter.

In a production model, replace the constant fraction with a sequence of stress assumptions. Let fund size, aggregate requests, and offer capacity vary at each step. If an analyst assumes improving acceptance, the report should identify the reason—such as a lower assumed tender volume—instead of embedding optimism in the formula.

Turn access uncertainty into a funding gap

Now assume the investor owes $600,000 on a fixed future date and has $200,000 in immediately available cash. The fund position is worth $1 million before any modeled valuation change. Assume a repurchase settles before the deadline; if it does not, its contribution to that deadline is zero.

Define the required additional funding as:

Here $L$ is the liability, $C$ available cash, $I$ other independently confirmed inflows, $V$ the tendered position's starting value, $m$ an assumed NAV decline, and $f$ a proportional exit fee. All monetary terms use the same currency. Set $I=0$ for this example.

The 10% decline and 1% fee are illustrative inputs, not claims about a particular product. Notice how access restrictions can dominate the near-term funding gap even without a valuation loss. Conversely, a full repurchase does not protect an investor from receiving a lower price.

For a binding obligation, the most important sensitivity may be timing. If settlement misses the deadline, the modeled gap becomes $400,000, regardless of an eventual $250,000 payment. Do not average late and timely proceeds together and call the result available liquidity.

Build a ledger that can survive scrutiny

The Financial Stability Board's May 6, 2026 private-credit report identifies valuation opacity, liquidity mismatches, and interconnected financing as vulnerabilities requiring attention. It also emphasizes limitations in available data. That supports testing assumptions rather than interpreting a stable reported NAV as evidence of frictionless exit. It does not establish that a particular fund is insolvent. FSB, Report on Vulnerabilities in Private Credit.

Our proposed implementation is a dated cash ledger with three separate views: contractual terms, modeled outcomes, and settled cash. Keeping those views distinct prevents a scenario from becoming an operational payment instruction.

1. Capture terms with provenance. Store the document version, share class, notice requirement, dealing cutoff, pricing date, payment deadline, repurchase amount, fees, and treatment of unfilled requests. Record ambiguous terms for review rather than selecting the most favorable interpretation.
2. Build the liability calendar. Enter payroll, capital calls, debt service, and other commitments by actual due date. Separate unavoidable obligations from discretionary spending. Convert currencies explicitly and keep FX assumptions visible.
3. Generate scenario paths. Apply different aggregate tender fractions, NAV paths, and settlement dates. Include a case with no repurchase cash before the relevant liability. Do not label a scenario “95% confidence” without a defensible statistical model and data.
4. Reconcile operational events. A submitted request is not an accepted request, and acceptance is not payment. Record confirmations and timestamps. Make the treasury forecast change state only when supporting evidence arrives.
5. Escalate the largest dated gap. Report the maximum cash shortfall across the modeled horizon, its date, and the assumptions causing it. Assign an owner to any funding response and check its availability before counting it as liquidity.

A minimum data record should include fund_id, share_class, document_version, request_id, requested_shares, accepted_shares, pricing_date, nav_per_share, fees, payment_due_date, and settled_amount. Unknown fields should remain unknown. Replacing missing acceptance data with the requested amount is a particularly dangerous default.

Interpret the results without overstating them

These calculations assess access to cash, not expected return or the fair value of the underlying loans and properties. They cannot determine whether an allocation is attractive, whether a manager will restrict withdrawals, or whether a secondary sale is available at a particular discount.

The repeated-offer model is intentionally simple. Real requests can rise or fall; NAV changes; distributions may provide cash; and a fund may have different mechanisms from those assumed here. Borrowing can alter liquidity but adds interest, covenants, and refinancing exposure. A facility should not enter the cash ledger merely because a lender might offer one.

Notice periods can give managers time to arrange orderly sales. Repurchase limits can reduce the need for immediate liquidation. Neither feature removes portfolio risk, and neither is inherently evidence of misconduct. The investor's task is to match the product's actual access conditions with the timing and certainty of their own obligations.

Architectural / Strategic Summary

• Model time and amount separately: notice, dealing, acceptance, pricing, and settlement each constrain usable cash.
• Translate aggregate caps into individual proceeds: in the example, a 5% offer against 20% tender demand accepts only 25% of a request.
• Avoid invented exit dates: repeated partial repurchases depend on future demand and the rules for resubmission.
• Stress the payment deadline: late proceeds contribute zero to an earlier obligation, even when the eventual payment is likely.
• Keep evidence states distinct: contractual terms, scenario cash, accepted proceeds, and settled money belong in separate fields.
• Treat the FCA development accurately: the October 8 consultation is a catalyst for planning, not an already enacted universal 90-day rule.