Financial Reporting
Nobody Makes You Disclose Your Valuation Policy. Your Auditor and the SEC Still Read It.
GAAP stopped making you disclose it, but your auditor and an SEC examiner still read your fund's valuation policy. What belongs in it before year-end marks.
By 409.AI Team - 2026-10-05
# Nobody Makes You Disclose Your Valuation Policy. Your Auditor and the SEC Still Read It.
Q3 closed last week. If you run finance at a first-time fund, the next sixty days decide how your first audit goes, and the deciding factor usually isn't the marks. It's a document most emerging managers write once, in a hurry, during fund formation, and never open again: the valuation policy.
It sits near the top of the audit request list. It's also what an SEC examiner asks for when they want to know whether your Level 3 numbers came from a process or from a conversation. And here's the part that surprises people: U.S. GAAP stopped requiring you to describe that process in your footnotes years ago. The obligation to have one didn't go anywhere.
The footnote that went away
ASC 820 used to make reporting entities disclose a description of the valuation processes behind Level 3 measurements. [ASU 2018-13](https://viewpoint.pwc.com/dt/us/en/fasb_financial_accou/asus_fulltext/2018/asu_201813fair_value/asu_201813fair_value_US/asu_201813fair_value_US.html) removed that requirement as part of the FASB's disclosure framework project, which weighed each disclosure's cost against what it actually told investors. What stayed, and got sharper, was everything quantifiable: the techniques, the significant unobservable inputs and their ranges, plus a narrative on measurement uncertainty.
So the policy went quiet. It stopped being something LPs read in the back of the financials and became an internal document, which is an easy kind of document to let rot.
Three parties still read it. Your auditor tests your marks against it. An SEC examiner reads it if you're registered. And in a Fund II diligence process, a sophisticated LP's operational due diligence team asks for it by name.
What the policy decides before the quarter does
The useful way to think about a valuation policy is that it makes decisions while nothing is at stake, so you're not making them in January with a 2.1x TVPI on the line.
It should pin down the unit of account for each kind of position, because that's what ASC 820 measures. For most venture funds that's the individual security held, not the enterprise. It should name a technique per asset type rather than one technique for everything: a backsolve or option-pricing allocation for a recently priced equity position, a market-multiple or income approach calibrated to the last transaction for an older one, and something explicit for instruments that aren't equity yet. If you hold SAFEs or convertible notes, say so and say how, because "cost until it converts" isn't a fair value measurement. We've written about [why a convertible note needs a with-and-without measurement rather than a placeholder](https://409.ai/articles/convertible-note-with-and-without-valuation-method).
Calibration belongs in there as a standing practice, not a one-time step. At initial recognition the transaction price is your reference point, and the model gets tuned to reproduce it so that later movements mean something. The SEC staff came back to calibration repeatedly in its September 28 statement, recommending "robust calibration practices, including periodic reassessment of whether model outputs remain consistent with available market information." Our walkthrough of [what that statement expects from your year-end marks](https://409.ai/articles/sec-staff-statement-private-asset-fair-value-calibration-nav) covers the measurement mechanics in detail.
The policy also has to name the triggers that force a remark: a new priced round, a secondary trade in the name, a down round or structured extension, a missed plan by some stated margin, a bridge from insiders, a change in the fair-value rules, or simply the passage of four quarters without a fresh data point. Triggers are what stop a position from drifting at cost for two years because nobody had a reason to touch it.
Three positions, one awkward quarter
Say you run a $40 million seed fund with 22 positions. Three of them show what the policy is for.
Northfield is easy. You put in $1.5 million five months ago at $3.00 per share in a priced Series A, so you hold 500,000 shares. The transaction price is your calibration point, nothing has changed, and the mark stays $1.5 million. Nobody argues about Northfield.
Harbor Logic is the real test. You invested $2.0 million at $4.00 per share, 500,000 shares, nineteen months ago. Revenue is up about 70% since then. Public comparables in the category have compressed. Your policy says a position without a transaction in four quarters gets remarked, and it says how: a market-multiple approach cross-checked against the allocation from the last round. Run it and you get $4.60 per share, so $2.3 million. A 15% markup on a company that grew revenue 70% feels stingy, and that reaction is exactly why the method needed to be chosen in advance.
Vantage Mill is the one most first funds get wrong. It's a $750,000 SAFE with a $12 million post-money cap, written 22 months ago, and the company just took a small insider bridge at the same cap. Carrying it at $750,000 because it hasn't converted isn't a measurement, it's an absence of one. The policy has to say what instrument you're holding and how its fair value responds when the only new evidence is a flat insider round.
Who marks, who reviews, who signs
A policy that says "the Investment Committee determines fair value quarterly" is not governance. It's a sentence.
Separate the roles, even in a three-person firm. Someone prepares the mark and assembles the support. Someone who didn't prepare it reviews the method and the inputs. Someone approves. Write down what happens when the reviewer and the deal partner disagree, because that conversation will happen on your most-marked-up name, and whoever wins it should not be deciding the process at the same time.
Then memorialize it. A valuation memo per position, quarterly minutes naming who attended and what changed, and a file holding the actual inputs: the comparable set with the date you pulled it, the cap table used in the allocation, the secondary quote and whether it was binding. Judgment that lives only in someone's head looks identical to no judgment at all once a year has passed.
Using an outside valuation specialist doesn't move the responsibility. The SEC staff was blunt about ownership of the number, noting that "a lack of timely information does not relieve management of its responsibility to estimate fair value." A specialist's report is evidence you considered; the mark is still yours.
The two-number problem
Here's the failure that draws attention fastest. Harbor Logic sits at $4.60 in your audited financials, and at $6.00 in the Fund II deck, because the deck used a term sheet that never closed.
That gap is the single most legible valuation problem a fund can have, because it needs no modeling to spot. One set of numbers persuades investors and another satisfies the auditor, and the policy either applies to both or it's decoration. If marks used in fundraising material come from a different process than marks in the financials, say so explicitly and be prepared to defend why. Most funds can't, which is the point.
Quarter-end evidence cuts the other way too. When a transaction prices below your carrying value, [the mark moves before your thesis does](https://409.ai/articles/continuation-fund-price-below-nav-asc-820-quarter-end-mark), and the policy should already say that an observed price in the name beats an internal model.
What the audit actually tests
AU-C 540, as rewritten by SAS No. 143 and effective for periods ending on or after December 15, 2023, frames the audit of an estimate around three things: the method, the significant assumptions, and the data. Your auditor evaluates each against what a reasonable preparer would do, and separately evaluates whether your disclosures about estimation uncertainty hold up.
The standard also requires a retrospective review. Last year's estimates get compared against what actually happened, not to grade your forecasting, but to detect bias in one direction. A fund whose marks are consistently late to move down has a pattern, and the pattern shows up in that review. This is the same lens a company's first audit applies to its 409A, where [the test is the method and the assumptions rather than the fee you paid](https://409.ai/articles/first-audit-409a-valuation-au-c-540-assumptions).
Classification gets tested too, and it's unforgiving: [one significant unobservable input sends the whole measurement to Level 3](https://409.ai/articles/asc-820-level-3-classification-significant-unobservable-input), with the disclosure load that follows. If your financials are prepared under ASC 946, the [investment company presentation rules](https://409.ai/articles/asc-946-investment-company-venture-fund-financial-statements) sit on top of all of this.
Why 2026 made this an exam question
The SEC's [2026 examination priorities](https://www.sec.gov/files/2026-exam-priorities.pdf) name two groups that describe most emerging managers: advisers to newly launched private funds, and advisers "that have not previously advised private funds," the latter reviewed for "regulatory awareness, liquidity, valuation, fees, disclosures, and differential treatment of investors, including use of side letters." Valuation is on that list by name.
The September 28 [staff statement on fair value for private assets](https://www.sec.gov/newsroom/speeches-statements/hohl-daley-statement-fair-value-measurement-disclosure-considerations-private-assets-092806) put it plainly: "robust policies and procedures, paired with material disclosure, help investors understand an entity's fair value process, the judgments involved, and the risks associated with private assets." It creates no new obligations. It describes what the staff expects to find.
If you're a registered adviser, Rule 206(4)-7 under the Investment Advisers Act already requires written policies and procedures and a review of their adequacy "no less frequently than annually." A valuation policy you haven't revisited since formation fails the second half of that sentence even if it passes the first.
Five things to write down before December
Pin the unit of account for each position type. Name the technique per asset class, SAFEs and notes included. List the events that force a remark, with the four-quarter stale rule as a backstop. Split preparer, reviewer and approver, and say how a disagreement resolves. Put the annual policy review on the calendar with a date and an owner.
Rule changes will keep arriving, and the policy is what absorbs them. [ASU 2026-03 on lock-up discounts](https://409.ai/articles/asu-2026-03-final-lock-up-discount-quarter-end-decision-funds) is adoptable now, which means your policy should say whether you've adopted it and from which quarter, not leave your auditor to infer it from the numbers. The broader mechanics of [marking a startup portfolio under ASC 820](https://409.ai/articles/asc-820-level-3-fair-value-fund-portfolio-valuation) are worth reading alongside it.
The policy isn't compliance paperwork sitting next to the real work of valuation. It's the part of the valuation that you can still write while you're objective. Every quarter you delay it, you're choosing to make those decisions later, with a number already in front of you and a reason to want it.
If the marks themselves are the bottleneck, having them prepared and reviewed by valuation specialists is what our [ASC 820 portfolio valuation](https://409.ai/products/asc-820) engagements handle.