Technical Article · Editorial Line 01 · Business Valuation
The last round price lies.
Why valuing venture capital is value allocation, not future prediction.
Technical Article · Editorial Line 01 · Business Valuation
Why valuing venture capital is value allocation, not future prediction.
A startup raises capital at a post-money of two hundred million reais. The next day, the founder's common share is worth a fraction of that. The headline figure and the figure that matters to the holder of the common share are not the same number, and the distance between the two is where almost all the technical difficulty of valuing early-stage companies resides.
The explanation is structural. The post-money is the implied value of the share class that was just issued, generally preferred, multiplied by the participation it now represents. It describes the price of a specific instrument, on a specific date, with specific economic rights. It does not describe the value of the company as a whole, much less the value of a common share that sits below in the liquidation waterfall, with no preference, no cumulative dividend, no anti-dilution protection. Treating the post-money as if it were the value of the company is the original error that contaminates a large portion of the portfolio valuations circulating in the market.
Best practice begins with an inversion of perspective. Valuing a venture capital investee has less to do with predicting which startup will scale than with consistency: given what the market just paid for an instrument, what is the total equity value that makes that price coherent, and how does that total value distribute among instruments with different rights. This is the backbone of the AICPA valuation guide for private equity and venture capital fund investments, the densest technical reference on the topic, and it is also what the December 2025 edition of the IPEV guidelines reinforces by placing calibration, complex capital structures, and hybrid instruments at the center of the discussion.
Calibration is the act of adjusting the valuation model so that it reproduces the price of an observed transaction on the date it occurred. When a company raises in a round with meaningful participation of independent investors, in conditions approximating an orderly transaction, that price is the best available evidence of fair value. The disciplined valuer calibrates the model to that point and, on subsequent dates, moves the valuation from there, updating what has changed in the business and in the market.
The common error appears a step later, in failing to return to the anchor after fixing it. The calibration performed at fundraising and then forgotten ages silently. The interest-rate environment changes, comparable-company multiples compress, the company burns cash faster than projected, and the model remains anchored to a reality that has passed. A mark that does not revisit calibration has the appearance of prudence and operates as inertia, locked to an outdated snapshot. The IPEV 2025 guidelines are explicit in treating calibration as a continuous process, not as a single event on the entry date.
The choice of method responds to the stage of the company and the quality of available information, not to the valuer's preference, and organizing reasoning by maturity prevents the valuation report from being reduced to applying formulas without understanding what they measure.
At the early stage, with no relevant revenue and no reliable comparables, the strongest evidence is the recent transaction itself. Here the backsolve dominates: the application of the option pricing model to solve for the total equity value that reconciles with the price paid for the most recent preferred share. The model treats each share class as an option on the company value, with strike prices defined by liquidation preferences, and uses the Black-Scholes framework to distribute value across classes. More heuristic methods, such as Bill Payne's Scorecard, the Berkus method, and the venture capital method described by Sahlman, have their place as sanity checks but are fragile as a single basis, because they depend on parameters that the valuer chooses without market anchoring.
At the growth stage, with recurring revenue and an observable trajectory, weight shifts to calibrated revenue multiples and to the tension between the market metric and the cash-burn reality. This is the terrain on which the valuation reset of the past few years manifests clearly: companies that raised at multiples of ten or fifteen times recurring revenue came to be worth half of that when the cost of capital rose, and the down round, far from being an anomaly, is calibration correcting itself. The option pricing model remains useful for allocating value across classes, but the estimate of total value rests on updated comparable evidence.
At the late stage, near a liquidity event, the company again behaves as an object of discounted cash flow and traditional multiples. Allocation between classes moves to the probability-weighted expected return method, known as PWERM, or the hybrid method, weighting scenarios of IPO, strategic sale, and dissolution. It is precisely in the weak-exit scenario that the liquidation preference, which appeared to be a legal detail, reveals itself as the dominant factor in the value of the common share, and the paradox at the opening of this article resolves: in a sale at a value close to invested capital, preferred shares receive first and the common share receives little or nothing.
A round-number example closes the argument. Suppose the latest round priced the preferred at a post-money of R$ 200 million, with R$ 80 million of non-participating liquidation preference ahead of the common share. In a sale at R$ 90 million, the preferred exercise the preference and withdraw R$ 80 million, leaving R$ 10 million for all the common shares that the post-money suggested were worth much more. In an IPO at R$ 600 million, the same preferred waives the preference, converts to common, and divides the value proportionally. The common share is worth one thing in one outcome and something very different in the other, and it is this optionality that the option model prices and that the headline figure ignores.
The 2025 edition of IPEV gives renewed attention to complex capital structures and hybrid instruments, and the reason is practical. The cap table of a mature startup is rarely a clean stack of shares. There are multiple series of preferred shares with different preferences, convertible instruments and SAFEs that have not yet converted, participation clauses, anti-dilution clauses, multiple preference clauses. Each of these terms alters the waterfall, and ignoring them in allocation produces a value of common share that appears precise and is simply wrong. Consistency requires that the allocation model reflect the actual contractual waterfall, not a convenient simplification.
The option pricing method has become the standard instrument for allocating value across share classes because it elegantly translates a correct intuition: each share class is an option on the value of the company, which only receives something above a certain threshold defined by the preferences of those ahead of it in the waterfall. The model requires four inputs, and each is a decision that must be substantiated: the total equity value, the expected volatility of that value, the time to liquidity event, and the risk-free rate. Volatility is the most delicate parameter, because there is no historical price series for a private company, and estimating it by reference to comparable listed companies requires judgment on leverage, stage, and sector.
The limit of the method lies in its central premise. The option pricing model assumes a single liquidity event in a defined horizon, and distributes value as if the company moved toward a single exit. The reality of an investee is branched: it may go public, may be sold to a strategic at a high value, may be sold in distress at a value close to invested capital, may have no exit at all. Each of these paths radically alters how much each share class receives, and this is precisely the branching that the probability-weighted expected return method captures by modeling discrete scenarios and weighting them. The hybrid method combines the two, using the option model within one or more PWERM scenarios. The choice between them has practical consequence: the more the capital structure and the array of exits become complex, the less defensible it becomes to rely solely on the option model with a single exit.
All the discipline described thus far presupposes something that is scarce in Brazil: frequent rounds, with independent investors, in conditions approximating an orderly transaction, that serve as a reliable calibration anchor. The local venture capital market is shallower and more concentrated, rounds are fewer, and information on terms is less transparent. When the latest relevant transaction is old, or was conducted in conditions not approximating market, the anchor loses force, and the valuer must acknowledge this rather than feigning a precision that the evidence does not support.
There is also the foreign-exchange layer. Many rounds of Brazilian startups are denominated in dollars, while reporting and accounting occur in reais, and instruments such as SAFEs and convertibles add a lag between cash received and shares to be issued in the future. Calibrating to a price in foreign currency, moving the valuation over time, and converting to the functional currency without double-counting FX risk requires a care that hasty work often skips. The most common temptation, and the most frequent error, is to import multiples from mature, deep markets and apply them to a Brazilian company without adjusting for differences in liquidity, cost of capital, and depth of the exit market. The imported multiple appears sophisticated and functions as a foreign anchor for a boat sailing in another sea.
The 2025 edition of the IPEV guidelines innovates by explicitly addressing the use of artificial intelligence in valuation processes. The tool enters under the same requirement as any other input, with traceability, substantiation, and human responsibility for the conclusion; prohibiting or endorsing the technology wholesale misses the point. A model that suggests a volatility, a comparable, or a scenario probability does not relieve the valuer of understanding where that number came from and why it holds. Automation accelerates the work but does not transfer technical responsibility, and a valuation that cannot explain its own inputs does not defend itself before an auditor, regardless of who or what generated them.
"Valuing venture capital has little to do with optimism or pessimism. It is, rather, disciplined allocation of a total value across distinct economic rights, anchored in market evidence and auditable step by step."
The discount rate, the multiple, the option model volatility, the probability of each exit scenario: none of these parameters is a number sought to reach a desired conclusion. Each is a consequence that is justified and documented. The AICPA guide and the IPEV guidelines exist so that this chain of justifications is consistent across valuers and over time, and it is in this consistency, not in the apparent sophistication of the mathematics, that the value of an independent valuation resides.
The headline figure will continue to be cited, and there is no problem with that. The valuer's job is to show, with method, what question that number answers and what question it does not.
References