Five Truths About Valuation — Why All Models Are Wrong but Some Are Useful

Damodaran's first truth is uncomfortable but essential: there is no such thing as an objective, unbiased valuation. Every valuation starts with someone's narrative about a company — and the numbers follow the narrative, not the other way around. The analyst covering a stock they want to initiate at Buy will estimate higher growth rates, more optimistic margins, and lower discount rates than an analyst who is short the same stock. The bias is not always conscious — it is often structural.

• Sources of structural bias: investment bankers are paid for completed transactions — they are incentivized to reach valuations that support the deal. Sell-side analysts earn revenue from trading commissions — positive coverage generates more transactions. Management teams presenting acquisition targets highlight synergies while minimizing integration costs. Even independent analysts anchoring on the current market price unconsciously reverse-engineer inputs to confirm it.
• The confirmation bias loop: most analysts build a DCF by starting with their investment view and adjusting inputs until the model confirms it. If you 'know' the stock should trade at $100, you will find assumptions that produce $100. This is the opposite of honest modeling, which would start with the best unbiased estimates of inputs and accept whatever value the model produces.
• Truth 2 is arithmetic: if valuation is uncertain (which it always is for any company with a future), then most point estimates will be wrong. The question is not whether you will be wrong — you will be — but whether you are wrong within a defensible range and wrong on the right side of the margin of safety.
• The practical implication: present valuations as ranges, not point estimates. 'This company is worth between $70 and $95 under reasonable assumptions, with my best estimate at $82' is honest. '$82 per share' is spuriously precise. Damodaran's templates always show a sensitivity table that makes uncertainty explicit.

Professional finance training produces analysts who believe more complex models are more accurate models. The evidence suggests the opposite. A 10-row DCF with explicit, defensible assumptions about a handful of key drivers typically outperforms a 500-row model that attempts to forecast every line item of the financial statements three years out. Why? Because the additional complexity does not add information — it adds opportunities for hidden errors and false precision to compound.

Damodaran's most profound insight about DCF modeling: every number in a financial model is the quantification of a narrative. The 25% revenue growth assumption means you believe this company will win market share at this rate. The 18% operating margin assumption means you believe these competitive advantages will be sustained at this level. Before a number goes into a model, there should be a clear statement of what business reality it represents. When analysts go straight to numbers without the narrative, they build models that are internally inconsistent — the growth assumption implies one competitive reality while the margin assumption implies another.

• The narrative-to-numbers process: start with the story. 'This is a platform business that will capture 15% of a $50B market over 10 years, generate SaaS-like margins above 30%, and require minimal physical capital.' Now translate each element into model inputs: revenue ceiling at $7.5B, margin trajectory to 30%+, low capex. The narrative makes each assumption testable and coherent.
• Cross-narrative consistency check: if you assume Apple will grow at 20% for 10 years, you are implicitly assuming Apple will be 6× its current size. Does that world exist? Does the market exist to support it? Will competitors allow it? Cross-checking the narrative's implications reveals which assumptions are implausible and need revision.
• Truth 5 — Markets are often right: the fifth truth is the most humbling. In the long run, stock prices do reflect fundamental values with reasonable accuracy. If your DCF says a company is worth $30 and the stock trades at $80, the more likely explanation is that your model is missing something important — not that the market is missing something important. Damodaran: 'The default assumption should be that the market knows something you don't, not that you know something the market doesn't.'
• Calibrating when to trust the market vs. your model: trust your model when (a) you have done the work carefully, (b) you can explain specifically what the market is getting wrong and why, and (c) the divergence is large enough to be meaningful. Distrust your model when (a) the divergence first emerged, (b) you cannot identify a specific error in the market's reasoning, or (c) every smart person you respect disagrees with your conclusion.

The five truths, taken together, define what disciplined valuation practice actually looks like versus what finance courses and textbooks often imply it looks like. A good valuation is not a spreadsheet that produces a precise number — it is a structured argument about a business's future, made quantitative, that acknowledges its own uncertainty: