Financial Reporting
ASC 820 and the Level 3 Problem: How Funds Mark Their Startup Positions at Fair Value
ASC 820 governs how VC and PE funds mark startup positions at fair value: Level 3 rules, calibration, backtesting, and the ASU 2022-03 lock-up change.
By 409.AI Team - 2026-07-20
# ASC 820 and the Level 3 Problem: How Funds Mark Their Startup Positions at Fair Value
It's audit season, and your fund's biggest position closed its last round eighteen months ago. The company is doing fine. Revenue is up, the team is intact, nobody's raising. Then your auditor asks the question that decides how the next three weeks of your life go: "Is that last-round price still fair value?"
If your only answer is "well, that's what the shares cost us," you're going to have a hard audit.
ASC 820 is the FASB standard that governs how venture capital and private equity funds measure and disclose the fair value of their portfolio investments. Most of what a fund holds are private, illiquid startup positions, and those sit at the bottom of the fair value hierarchy, where judgment does the heavy lifting and auditors ask the most questions. Here's what the standard actually requires, where funds get tripped up, and what changed recently that affects the marks you're signing off on this year.
What ASC 820 actually asks of you
ASC 820 defines fair value as the price you'd receive to sell an asset in an orderly transaction between market participants at the measurement date. That's an exit price, not what you paid, and not what you hope to get someday. It's a hypothetical sale, today, to a willing and informed buyer.
Three words in that definition carry more weight than they look like they should. "Orderly" rules out a fire sale. "Market participants" means you value the position the way a typical buyer would, not the way you, with your particular tax situation or portfolio, happen to see it. And "measurement date" means the number reflects what's known as of the reporting date, not a stale figure from the last financing.
This is a different exercise from a [409A valuation](https://409.ai/articles/409a-valuation-vs-fair-market-value), which sets the strike price for a portfolio company's own employee options. A 409A values one company's common stock for tax compliance. ASC 820 values your fund's holding, often the preferred stock, for financial reporting to your LPs. Same company, different question, frequently different number. Conflating the two is one of the faster ways to draw an audit comment.
The fair value hierarchy, and why startups land at Level 3
ASC 820 sorts the inputs you use into three levels based on how observable they are.
Level 1 is a quoted price in an active market for an identical asset. Think of a public stock with a closing price on the measurement date. There's nothing to estimate.
Level 2 uses observable inputs other than a Level 1 quote: prices for similar assets, or quoted prices in markets that aren't especially active. You're leaning on real market data, just not a direct quote for the exact thing you hold.
Level 3 uses unobservable inputs. You're building an estimate from internal assumptions and models because the market isn't handing you a price. This is where nearly every private startup position lives. There's no ticker for a Series B stake in a company that raised eighteen months ago, so you construct fair value from what you can support.
The level isn't a grade on your work. It describes the inputs, not the quality. But Level 3 carries the most disclosure and the most scrutiny, because the number depends on assumptions only you can see. That's precisely why your documentation has to be airtight.
The three approaches, and picking the right one
ASC 820 recognizes three valuation approaches, and the [AICPA's Accounting and Valuation Guide for portfolio company investments](https://www.aicpa-cima.com/resources/article/valuation-of-portfolio-company-investments-of-venture-capital-and-private) walks through how to apply each one to Level 3 holdings.
The [market approach](https://409.ai/articles/market-approach-409a-valuation) prices your position off comparable transactions or trading multiples. For a recently funded startup, the most recent round often anchors this, at least at first.
The [income approach](https://409.ai/articles/income-approach-409a-valuation) discounts the company's expected future cash flows back to present value. It's more work and demands defensible projections, which is why it's used more for companies with real revenue than for a pre-product seed bet.
The [cost, or asset-based, approach](https://409.ai/articles/asset-based-approach-409a-valuation) builds value from the company's underlying net assets. For an asset-light software startup it usually understates fair value, so it tends to serve as a floor or a cross-check rather than the primary method.
Most funds use the market approach for the bulk of their book, then reach for the income approach as portfolio companies mature and generate cash flows worth discounting.
"Just mark it at the last round" is where audits go sideways
Here's the trap. A recent financing is strong evidence of fair value on the day it closes. An arm's-length round with new outside investors is close to a real transaction price, which is about as good as Level 3 evidence gets. So marking to the round right after it closes is defensible.
The problem is time and the assumption that the round price is frozen in place. Fair value is measured at each reporting date, so the question isn't "what did the last round imply?" It's "what would a market participant pay today?" A lot can move between the two. The company might have blown past plan, or missed it badly, or the comparable public companies you'd benchmark against might have re-rated by 40% since the round.
This is the same disconnect founders hit from the other direction when they learn [why a 409A comes in below the post-money valuation](https://409.ai/articles/why-is-your-409a-valuation-lower-than-post-money-valuation): the headline round number reflects the rights of the newest preferred shares, not a uniform per-share value across the whole cap table. A round price is a starting point, not a permanent answer.
Calibration and backtesting: the discipline auditors look for
Two techniques from the AICPA guide separate a clean audit from a painful one.
Calibration means anchoring your valuation model to the transaction price when you first buy in, then holding onto that model. If you paid a price that implied, say, a 6x revenue multiple, you calibrate your model so it reproduces that entry price. At each later reporting date, you update the inputs, current revenue, current comparable multiples, company progress, and let the calibrated model produce the new mark. You're not re-guessing from scratch each quarter. You're moving a well-anchored model forward as facts change. As a portfolio company hits its milestones, it may earn a multiple closer to its stronger peers, and calibration gives you a disciplined way to reflect that.
Backtesting means comparing your prior fair value estimates against what actually happened at the next real event, a follow-on round, a sale, a write-off. If your marks are consistently well below where the next round prices, or consistently above, that's a signal your process needs adjusting. Auditors ask for this. Coming to the audit with your own backtesting already done is far better than having them discover the pattern for you.
What changed: lock-ups and ASU 2022-03
If your fund holds shares subject to a contractual restriction on selling, a lock-up after an IPO, for instance, a recent standard changes how you treat it.
FASB's ASU 2022-03 clarified that a contractual sale restriction is a characteristic of the holder, not of the security itself, so it isn't factored into the fair value measurement under ASC 820. In practice, that means you generally can't take a separate discount just because you're contractually barred from selling for a lock-up period. Before the update, practice varied, and some funds applied a discount for that restriction while others didn't. The clarification removes that inconsistency, and it adds disclosure requirements around securities subject to those restrictions.
The timing matters for this year's reporting. The update is effective for private companies for annual periods beginning after December 15, 2024, so it's now in force for many funds' current financial statements. Public business entities have been applying it since annual periods beginning after December 15, 2023. If your prior marks quietly included a lock-up discount, this is the year to revisit them. (See KPMG's summary in its [Financial Reporting View library](https://kpmg.com/us/en/frv/reference-library/2022/fasb-clarifies-fair-value-guidance-for-sale-restrictions.html) for the detail.)
Complex cap tables: not every share is worth the headline price
One more place funds get the number wrong: treating every security in a company as if it's worth the same per-share price the last round implied. It usually isn't. Preferred shares carry liquidation preferences, participation rights, and seniority that common shares don't, so their fair values diverge, sometimes sharply, especially in a down or sideways scenario.
Valuation specialists handle this with allocation methods like an option pricing model or a probability-weighted expected return method, which distribute a company's total equity value across share classes according to their rights. Instruments like [SAFEs add another layer](https://409.ai/articles/how-safes-affect-your-409a-valuation), since an unconverted SAFE sits in a gray zone between debt and equity until the trigger event that converts it. If you hold a mix of preferred, common, and convertible instruments in the same company, they generally shouldn't all carry the same mark.
Getting the audit right
ASC 820 rewards process over cleverness. The funds that sail through audits tend to do the same handful of things: they calibrate to entry price and carry the model forward, they document why they chose an approach and what inputs drove it, they backtest their own marks, and they bring in an independent valuation specialist for the positions that matter most. That last point isn't only about accuracy. An independent, well-supported valuation is far harder for an auditor, or an LP, to push back on than a number that came from the same desk that holds the position.
This is the same fair value discipline that shows up across financial reporting, from [purchase price allocation under ASC 805](https://409.ai/articles/asc-805-purchase-price-allocation-startup-acquisition) after an acquisition to [expensing stock options under ASC 718](https://409.ai/articles/asc-718-stock-based-compensation-startup-guide). The standards differ, but the core question, what would a market participant pay, is the same one every time.
If you manage a fund and your portfolio marks are heading into an audit, the worst time to build your valuation support is after the auditor's first question. 409.ai produces expert-reviewed [ASC 820 fair value reports](https://409.ai/products/asc-820) for fund portfolios, built to hold up to LP and audit scrutiny. Get the calibration file, the approach memo, and the backtesting in place before the audit starts, and the question about your eighteen-month-old round becomes a short conversation instead of a three-week one.