Table of Contents
The finance industry loves to present itself as a bastion of mathematical certainty. Walk into any investment bank and you’ll find analysts hunched over spreadsheets, their screens glowing with numbers that stretch into the future. They’re building discounted cash flow models, or DCFs, which are supposed to tell us what a company is truly worth. The models look scientific. Valuations feel objective. And yet, ask two analysts to value the same company and you’ll get two different answers. Sometimes wildly different answers.
This isn’t a bug in the system. It’s the system itself.
The Illusion of Objectivity
The DCF model is deceptively simple in concept. You predict how much cash a company will generate in the future, then discount those cash flows back to the present using a rate that reflects risk and the time value of money. Add it all up, and you get a number. That number is supposed to be the intrinsic value of the company. But tucked inside this elegant framework are dozens of assumptions, each one a trapdoor through which objectivity can escape.
Consider the growth rate. Every DCF model needs one. How fast will this company’s revenues grow over the next five years? Over the next ten? The analyst must decide. They’ll look at historical performance, industry trends, management guidance, and competitive dynamics. They’ll consider macroeconomic factors and technological disruption. And then they’ll pick a number. Another analyst, looking at the exact same information, might pick a different number. Not because one is incompetent, but because they weight the evidence differently. They tell themselves a different story about the future.
This is where valuation starts to resemble art more than science. A painter and a photographer might both capture the same landscape, but their final works will look nothing alike. Each brings their own perspective, their own interpretation of what matters. The painter might emphasize the emotion of a stormy sky. The photographer might focus on the sharp detail of a single tree. Neither is wrong. Both are incomplete.
The Discount Rate Dilemma
The discount rate introduces another layer of subjectivity. This rate is meant to reflect the riskiness of those future cash flows. A stable utility company gets a lower rate than a speculative biotech startup. But how much lower? The conventional approach uses something called the weighted average cost of capital, which sounds technical enough to be objective. It isn’t. Buried inside this calculation are estimates of market risk premiums, beta coefficients, and capital structure targets. Each estimate requires judgment. Each judgment creates room for disagreement.
Here’s the counterintuitive part. The most sophisticated analysts often produce the most divergent valuations. You might expect that more training and better tools would lead to convergence, to some shared understanding of value. Instead, expertise tends to amplify differences. A novice analyst might stick close to simple rules of thumb and industry averages. An experienced analyst knows enough to question those rules, to build more nuanced models, to see complexity where others see simplicity. They’ve learned which assumptions really matter, which means they’ve learned where to inject their own views most forcefully.
The Terminal Value Problem
The terminal value calculation is perhaps the best illustration of valuation as art. After projecting cash flows for five or ten years, the analyst must estimate what the company will be worth at the end of that period. This terminal value often represents more than half of the total valuation. Sometimes much more than half. Think about that for a moment. Most of the company’s value comes from a guess about what it will be worth in a decade, which itself depends on assumptions about growth rates and market conditions that might as well be written in sand.
The standard approach assumes the company will grow at some steady rate forever. Forever is a long time. It’s longer than any company has existed. It’s longer than most industries have existed. And yet, we type a number into a cell and pretend we’ve solved for eternity. One analyst might use 2% perpetual growth. Another might use 3%. That single percentage point difference, compounded into infinity, can swing the valuation by billions of dollars. Is the 3% analyst wrong? Not necessarily. They might believe the company has durable competitive advantages that will sustain above average growth. They might have more faith in management’s ability to reinvest capital productively. Or they might just be more optimistic by nature.
The Psychology Behind the Numbers
This brings us to an uncomfortable truth about valuation. It’s not just a technical exercise. It’s psychological. Two analysts can have identical training, identical information, and identical tools, and still arrive at different values because they see the world differently. One analyst might have lived through the dot com crash and learned to distrust high growth stories. Another might have started their career during a raging bull market and learned that skeptics miss out. These experiences don’t invalidate their analysis. They are their analysis, woven into every assumption.
The connection to other fields becomes obvious when you think about it this way. Literary critics reading the same novel will emphasize different themes and reach different interpretations. Historians examining the same event will construct different narratives based on which sources they trust and which questions they think matter. Doctors reviewing the same symptoms might recommend different treatments depending on their specialty and experience.
In each case, expertise doesn’t eliminate subjectivity. It channels it in more sophisticated directions.
Even the choice of which DCF model to use involves judgment. Should you discount free cash flows to the firm or free cash flows to equity? Should you use nominal or real cash flows? How should you account for options or convertible securities? These aren’t details. They’re fundamental decisions that shape the entire analysis. The analyst must choose, and in choosing, they reveal something about how they understand value creation.
The market seems to know this better than academics do. If DCF models produced objective valuations, stock prices would converge to those values. Mispricings would be rare and short lived. Instead, stocks trade at premiums and discounts to their calculated intrinsic values all the time. Sometimes for years. Either the models are consistently wrong, or value itself is slipperier than we’d like to admit.
Both explanations point in the same direction. Valuation is interpretation.
Why This Matters
This doesn’t mean valuation is useless. A painting isn’t useless just because it reflects the artist’s perspective. It’s useful precisely because it does. The DCF framework forces the analyst to articulate a specific view of the future. It makes assumptions explicit rather than leaving them fuzzy. When two analysts get different answers, the interesting question isn’t who is right. It’s why they differ. What does Analyst A believe about the business that Analyst B doesn’t? Where do their mental models diverge?
The best investors understand this intuitively. They don’t treat valuations as facts to be discovered. They treat them as arguments to be constructed and then stress tested. They know their DCF model is wrong. The question is whether it’s useful. Does it help them think clearly about the business? Does it reveal which assumptions drive the value? Does it identify where they might be fooling themselves?
The Problem With Precision
This is why experienced analysts often become less confident in their precision even as they become more confident in their judgment. A junior analyst might proudly declare that a company is worth $47.23 per share. A senior analyst might say it’s worth somewhere between $35 and $55, and here’s what would have to be true for it to be worth more than $60. The junior analyst has false precision. The senior analyst has useful humility.
The false precision problem runs deep in finance. We dress up uncertainty in mathematical clothing and convince ourselves we’ve made it go away. The DCF model is particularly vulnerable to this because it produces a single number. The spreadsheet doesn’t output a range or a probability distribution. It outputs $47.23. That specificity is comforting. It’s also misleading. The appearance of precision makes us forget that we’re forecasting cash flows a decade into the future for a company operating in a changing industry facing unknowable competition.
Stories Disguised as Numbers
Consider what we’re really doing when we build these models. We’re telling stories about the future and then translating those stories into numbers. The translation creates an illusion of objectivity. Numbers feel harder than narratives. But the numbers are only as good as the narratives that generate them. If you believe the company will successfully expand into new markets, you’ll forecast higher revenue growth. If you doubt management’s capital allocation skills, you’ll use lower returns on invested capital. The model doesn’t make these judgments for you. You make them and the model dutifully calculates the implications.
This narrative dimension of valuation explains why two analysts can look at the same company and see completely different businesses. One analyst might see a dominant platform with network effects that will compound value for decades. Another might see a mature business facing commoditization and margin pressure. They’re not looking at different data. They’re constructing different stories from the same data.
Their DCF models simply formalize those stories in financial language.
The practical implication is that valuation should be iterative and collaborative. A single analyst working alone is likely to lock into their initial story and build a model that confirms it. They’ll pick assumptions that feel reasonable given their narrative, and the model will output a valuation that reinforces their view. This isn’t deliberate bias. It’s how human cognition works. We seek consistency. We smooth over contradictions. We convince ourselves that our assumptions are obvious and natural when they’re actually contingent and contestable.
Working with other analysts forces you to defend your assumptions. Why did you choose a 12% discount rate instead of 10%? Why are you modeling three years of high growth instead of two? Why did you assume margins will expand when the industry trend is flat?
These questions don’t have definitive answers, but the process of articulating answers helps. It surfaces implicit beliefs. It reveals which assumptions you’re confident about and which ones you’re guessing at. It turns valuation from a mechanical exercise into a genuine inquiry.
The Nature of Disagreement
The best valuation debates aren’t about who got the math right. They’re about whose story makes more sense. And stories can’t be right or wrong in the way that arithmetic can. They can only be more or less plausible, more or less consistent with the evidence, more or less useful for making decisions. Two plausible stories can coexist. This is why smart people disagree about valuations. They’re not failing at analysis. They’re succeeding at thinking independently.
The market’s function, in this view, is not to discover some true intrinsic value that exists platonically somewhere. It’s to aggregate different stories and different valuations into a price. That price represents a rough consensus, a weighted average of competing narratives. Sometimes the consensus is wildly wrong. That’s not a market failure. It’s a feature of trying to value uncertain future cash flows.
The wisdom of crowds works better for estimating the number of jellybeans in a jar than for predicting the future of a business anyway.
Embracing the Uncertainty
This is ultimately why valuation remains as much art as science. Science progresses by reducing disagreement. As we learn more, our theories converge. Our measurements become more precise. Our predictions become more accurate. But valuation doesn’t work this way. Better tools and more data haven’t made analysts agree more. If anything, they’ve enabled more sophisticated disagreement. The proliferation of valuation models, factor models, and scenario analyses has given analysts more ways to express their views, not fewer.
And maybe that’s how it should be. A business is not a natural phenomenon governed by fixed laws. It’s a human enterprise operating in a complex adaptive system. Its future depends on decisions that haven’t been made yet, by people who might not work there yet, in response to circumstances that haven’t occurred yet. Trying to capture that in a single number is audacious. The fact that we try at all, and that our attempts are sometimes useful, is remarkable.
So when two analysts produce two different DCF valuations, they’re both right. They’re right that valuation requires making assumptions. They’re right that reasonable people can differ on those assumptions. They’re right that thinking rigorously about value, even if you can’t pin it down precisely, beats not thinking about it at all. The art isn’t in eliminating disagreement.
The art is in understanding why you disagree, what that reveals about your worldview, and whether the story you’re telling yourself is the story you should believe.