Most people learn valuation as a checklist: DCF for "fundamentals," multiples for "markets," NAV for "assets." They pick a model, plug in numbers, and get the output value. This framing is wrong and its why smart analysts can value the same company and disagree by 50% while both being intellectually rigorous.

In 2021, analysts valued Rivian between $60 billion and $150 billion using the same public information. Tesla’s valuation swung wildly for years as investors disagreed not about the math, but about what kind of company it actually was. SoftBank’s Vision Fund deployed tens of billions using growth multiples that assumed sustained exponential expansion and suffered when those assumptions met reality.

These are not calculation errors. They're worldview mismatches.

Valuation models are not interchangeable tools. Each encodes beliefs about how the world works—about market efficiency, institutional trust, uncertainty, and time. Apply the wrong model to the wrong context and the output may look precise while being intellectually meaningless.

This blog examines what each major valuation framework actually assumes about reality, when those assumptions hold, and when they catastrophically fail.

1. Discounted Cash Flow: Valuation as Forecasting

The Discounted Cash Flow (DCF) model rests on a simple premise: a business is worth the present value of the cash it will generate in the future. At its best, DCF forces intellectual discipline. You must articulate explicit assumptions about revenue growth, margins, reinvestment needs, and risk. It separates operating performance from financing choices. It asks a hard question: what must this business earn to justify today's price?

But DCF has a dirty secret that most practitioners quietly ignore: the terminal value problem.

Consider a typical DCF model with a 10-year explicit forecast period. In most cases, especially for growth companies, over 70% of total enterprise value comes from the terminal value: a single assumption about perpetual growth beyond year 10.

Think about what this means. You're pretending to forecast cash flows ten years out, which is already heroic. Then you're basing most of your valuation on what happens in year 11, 15, 20, forever. The model doesn't reduce uncertainty—it just pushes it into a cell labelled "terminal growth rate" that determines everything but receives minimal scrutiny.

Change that terminal growth assumption from 3% to 4%, and equity value might swing by 40%. The entire valuation hinges on whether you believe sustainable long-run growth is 3.0% or 3.2%, a distinction no one can credibly make.

When DCF actually works:

DCF shines for mature, stable businesses in predictable institutional environments. Valuing Coca-Cola, Johnson & Johnson, or P&G makes sense via DCF because:

• •Cash flows are relatively stable and predictable
• •Competitive advantages are durable
• •Institutional environment (contracts, property rights, regulation) is stable
• •Terminal value is a smaller portion of total value because current cash flows are substantial

When DCF catastrophically fails:

Lehman Brothers in 2007. DCF models valued the firm based on projected investment banking fees and trading profits. They assumed continued access to short-term funding markets. When that access disappeared in September 2008, the DCF was irrelevant. The business didn't fail because cash flow projections were wrong—it failed because funding markets froze, and the institutional environment changed overnight.

WeWork in 2019. SoftBank's DCF assumed WeWork would achieve positive unit economics at scale, that the "community-adjusted EBITDA" metric meant something, and that real estate markets would accommodate infinite expansion. None were true. The model worked mathematically; the worldview was fantasy.

The real lesson: A DCF doesn't reveal truth. It reveals how sensitive your conviction is to assumptions you cannot possibly know. It's a framework for organizing uncertainty, not eliminating it. When institutional stability, funding markets, or business model viability is questionable, DCF becomes an exercise in false precision.

2. Relative Valuation: Valuation as Social Comparison

Relative valuation answers a fundamentally different question: How is this asset priced relative to others the market already values? Price-to-earnings ratios, EV/EBITDA multiples, price-to-sales. These compress information about growth, risk, profitability, and capital structure into a single comparable number. They're fast, intuitive, and anchored in market reality rather than your personal forecasts.

But they're also profoundly circular.

The 1999-2000 dot-com bubble is the canonical example. In early 2000, Yahoo traded at 700x earnings. Amazon at 1,500x earnings. Priceline.com at effectively infinite P/E because it had no earnings. An analyst comparing Pets.com to other e-commerce companies would conclude it looked "cheap"—trading at only 20x sales while comparable companies traded at 30-40x sales. The relative valuation was mathematically correct. The entire peer group was catastrophically overvalued. When Pets.com went bankrupt nine months after IPO, the relative valuation didn't "fail"—it never tried to answer whether the business model made economic sense. It only said: given that the market values unprofitable e-commerce companies at absurd multiples, this one is relatively cheap.

The 2021 SPAC bubble followed the same pattern. EV SPACs traded at 20-30x forward revenue projections. Analysts comparing new EV SPACs to existing ones concluded many looked "attractive" on a relative basis. Nikola, Lordstown, Faraday Future, Canoo—all looked reasonable relative to each other. But the entire sector was mispriced. Relative valuation just redistributed the error across companies.

When relative valuation actually works:

Relative valuation is powerful when you trust that the market is roughly efficient within a sector, even if you're agnostic about absolute levels.

Comparing two nearly identical regional banks during normal times tells you something real: the market prefers one over the other, probably for good reason (better loan quality, management, deposit franchise).

Comparing Pepsi to Coca-Cola using P/E ratios is informative because they're genuinely similar businesses facing similar economics.

What relative valuation reveals:

• •Who the market prefers and why
• •Where capital is flowing
• •Sentiment and market regime
• •Trading opportunities if you believe mispricing is temporary

What it does not reveal:

• •Whether the entire sector is mispriced
• •Whether returns will exceed cost of capital
• •Whether the business model is economically viable

Relative valuation tells you who is cheap relative to the crowd. It cannot tell you whether the crowd is wrong.

3. Asset-Based Valuation: Valuation When Trust Breaks Down

Asset-based valuation asks: what does the firm own, minus what it owes?

Book value, adjusted net asset value (NAV), and liquidation value dominate this framework. It's fundamentally pessimistic and it assumes you cannot trust management's ability to generate returns, so you focus on what could be recovered if everything stopped.

This approach dominates in three contexts:

Financial institutions: Banks are leveraged portfolios of financial assets. Their value depends on whether loans will be repaid and whether assets are marked accurately. During the 2008 financial crisis, DCF models of Citigroup or Bank of America were useless because no one knew what the mortgage portfolios were actually worth. Investors focused on tangible book value and regulatory capital ratios—what's definitely there versus what management claims.

Distressed companies: When a business is haemorrhaging cash and bankruptcy is possible, asset-based valuation dominates. What can secured creditors recover? What are the hard assets worth in liquidation? The business as a going concern has negative value, but the pieces might be worth something.

Emerging markets with weak governance: When you don't trust that management will distribute profits to minority shareholders, asset-based valuation provides a floor.

When asset-based valuation fails spectacularly: Applying it to platform businesses, software companies, or any business whose value lies in intangibles. Facebook in 2012 had minimal tangible assets—some servers, office furniture, maybe $15 billion in book value. Its market cap was $100 billion because its value was the network of 1 billion users, the data, and the advertising algorithms. Book value was irrelevant. Valuing Amazon by its warehouses and inventory misses that the value is in AWS, the logistics network, Prime membership, and merchant relationships—none of which appear meaningfully on the balance sheet.

The insight: Asset-based valuation only matters when future cash flows are so uncertain, or governance is so poor that you focus on what can be liquidated rather than what can be earned. It reveals institutional trust. If you're using NAV, you're admitting you don't trust management or the business model.

4. Sum of the Parts: Valuation as Intellectual Honesty

Conglomerates, financial groups, and hybrid business models break single-model valuation.

General Electric in 2015 combined industrial manufacturing, financial services (GE Capital), and media (NBC Universal). Applying a single P/E multiple to this mix was intellectually dishonest—each segment had completely different economics, risk profiles, and appropriate valuation frameworks. GE Capital should have been valued like a bank (price-to-book, return on equity). The industrial businesses should have been valued on EBITDA multiples or DCF. NBC should have been valued like other media companies. Instead, analysts often applied a blended multiple that obscured the fact that GE Capital was a leveraged black box, and the industrial margins were deteriorating.

The most common SOTP mistake:

Applying the same multiple across structurally different businesses because it's convenient.

Valuing Alphabet by applying a 30x P/E to the entire company misses that:

• •Search advertising is a mature cash cow deserving a lower multiple
• •YouTube is a high-growth media business deserving a higher multiple
• •Cloud is a capital-intensive infrastructure business with different economics
• •Waymo is an unprofitable R&D option

SOTP forces disaggregation. It's not a technique for managing complexity, it's a technique for intellectual honesty.

5. Market-Implied Valuation: Letting Price Speak

Instead of asking "What is this worth?", market-implied valuation asks: What must be true for the current price to make sense? This reverses the traditional valuation process. Rather than building a model and comparing it to the market price, you take the market price as given and reverse-engineer the assumptions.

Example: Tesla in 2020

Tesla traded at $400 billion market cap. Instead of building a DCF with your assumptions, run a reverse DCF: What revenue growth, margins, and scale must Tesla achieve to justify $400 billion?

The answer: Tesla would need to sell several million vehicles annually at operating margins higher than traditional automakers for decades—essentially becoming larger than Toyota while maintaining software-like profitability. That's not impossible, but it makes the embedded bet explicit. Bulls believe it. Bears don't. But everyone now knows what the price assumes.

Why this matters:

It shifts valuation from storytelling to falsification. You're not selling a narrative—you're exposing what the market already believes and asking whether that belief is defensible.

The best investors don't look for cheap assets. They look for mispriced expectations. Warren Buffett buying Coca-Cola in 1988 wasn't finding a "cheap" stock by traditional metrics. He was identifying that the market's expectations for international growth and pricing power were too conservative.

Market-implied valuation reveals hidden assumptions. Then you decide whether those assumptions are too optimistic or too pessimistic.

The Real Point

Valuation is not about choosing the "best" model. It's about choosing the model whose assumptions best match reality.

Intellectual failures, not technical ones:

Using a DCF in Venezuela assumes institutional stability that doesn't exist.

Using P/E multiples on banks in 2008 without understanding balance sheet risk assumes assets are marked accurately—they weren't.

Using book value on Facebook assumes intangible assets don't matter—they're everything.

These aren't spreadsheet mistakes. They're conceptual failures about what the model requires to be meaningful.

The choice reveals your worldview:

• •DCF = I trust my forecasts more than the market, and I believe institutional stability will persist
• •Multiples = I trust the market's collective wisdom about relative value, even if absolute levels are uncertain
• •Asset-based = I don't trust management or the business model, so I focus on liquidation value
• •SOTP = I believe different businesses within the same firm obey different economic logics
• •Market-implied = I want to know what the market believes, not what I believe

This is not an exhaustive list. There are many other ways to value a firm but they all follow the same principle: every method embeds assumptions about what drives value and what can be known with confidence.

Valuation is not objective. It cannot be. Every model encodes assumptions about trust, time, institutions, and uncertainty. The question is not which model is "right." The question is: which model's assumptions best describe the reality you're analysing?

Choose carefully. Your valuation reveals not just what you think an asset is worth—it reveals what you believe about how the world works.