Discounted Cash Flow, Without the False Precision

What DCF Actually Does

The standard explanation of DCF goes like this: project the company's future cash flows, add a terminal value, discount everything back to today, divide by shares outstanding, and you have the stock's intrinsic value. That framing is technically correct and practically misleading. It makes DCF sound like a calculation that produces a fact. It doesn't.

A DCF is only as reliable as the three sets of estimates you put into it. None of those estimates can be derived from the financial statements. They require judgment about the future, and two reasonable people with identical access to a company's filings will reach meaningfully different numbers. That's not a flaw in the tool. It's what the tool does.

The more useful framing: DCF is a structured way to ask what you'd have to believe for the current stock price to make sense. Run it backward from the market price. If a stock at $50 implies 14% free cash flow growth for ten years at a 10% discount rate, you can decide whether you believe that — and how wrong it would have to be before you lose money. DCF gives you the question. You supply the judgment.

The Three Numbers That Drive the Whole Thing

Three inputs do most of the work, and the order in which you build them matters because each one sets up the next.

Free cash flow is what you're forecasting. Take operating cash flow from the cash flow statement, subtract capital expenditures, and you have the standard version. Both line items appear in every 10-K and 10-Q. Some analysts adjust for stock-based compensation (it's non-cash but real dilution), normalize for unusual working capital swings, or strip one-time items. The adjustments can be legitimate — just be consistent and document why you made them. The goal is a number that represents what the business actually generates in cash after maintaining and growing its operations, before any financing decisions.

The discount rate is your opportunity cost — the return you're giving up by owning this stock instead of something else. Convert future dollars into today's dollars by dividing each year's cash flow by (1 + rate) raised to the power of that year. The formal version is WACC, which blends after-tax cost of debt with cost of equity, weighted by capital structure. WACC sounds rigorous. In practice, estimating cost of equity requires a market risk premium (a number academics have argued about for decades) and a beta (which is backward-looking, volatile, and measures price volatility rather than business risk). For most retail investors, using a round rate — 10% as a baseline, higher for riskier businesses — is more honest than false precision from WACC inputs that are themselves uncertain.

Terminal value is the third input and the largest single driver of your output. After your five- or ten-year explicit forecast, the terminal value captures everything after that. The formula is FCF₅ × (1 + g) ÷ (r − g), where g is the terminal growth rate and r is the discount rate. The trouble is that g has to be a number you make up — a guess about how fast the company grows from year 10 through infinity.

Terminal Value Is Where DCF Lives or Dies

Textbooks treat terminal value as step four in a five-step process — something you tack on at the end and then discount back. In practice it's the whole game. For a typical growing company, the present value of years one through five or one through ten accounts for only 20–40% of total estimated intrinsic value. Terminal value makes up the rest. Run the numbers on any mid-cap industrial, software company, or consumer brand with stable cash flows, and terminal value will represent 65–75% of the total. Occasionally more. The pull quote at the top of this page isn't rhetorical — 60–80% is the normal range, not an edge case.

Here's the arithmetic behind why that matters so much. The terminal value formula is FCF × (1 + g) ÷ (r − g). At a 10% discount rate and a 2% terminal growth rate, the denominator is 8%, and terminal value is roughly 12.75× year-5 FCF. Shift the terminal growth rate up by a single percentage point to 3%, and the denominator drops to 7%. Terminal value is now about 14.7× year-5 FCF. That's a 15% increase in terminal value from a one-point change in g — and because terminal value represents 70% of total estimated value, the intrinsic value per share goes up by roughly 10.5%.

A 0.5% change — not one full percentage point, half a point — produces something in the range of 10–20% swing in total intrinsic value depending on where you started. For a stock at $40, that's a $4–8 movement from a rounding decision on a number no one can know with certainty. The gap retail investors agonize over — is this worth $38 or $42? — is entirely inside the error bars of a half-point shift in terminal growth.

This is not a reason to avoid building DCFs. It's a reason to report your output as a range rather than a single number. A real DCF output looks like this: at a 1.5% terminal growth rate, fair value is $28. At 2.5%, it's $36. At 3.5%, it's $48. Those three numbers — not just the middle one — are the output of the model. The range tells you how sensitive your conclusion is to the one input you're least certain about.

Two analysts can run honest models on the same company with the same near-term FCF projections and produce intrinsic value estimates that differ by 25–30% purely because they chose different terminal growth assumptions. Neither has made an error. They have different views on a number that cannot be derived from first principles. When you hear two Wall Street analysts arguing about the same stock and reaching opposite conclusions, it's almost always this — not a disagreement about this quarter's results but a disagreement about what the company's long-run growth rate is, expressed as a number that moves the entire model.

Picking a Discount Rate Without Lying to Yourself

The discount rate represents your opportunity cost. It's the return you're requiring from this investment — and equivalently, the return you're forgoing by owning this stock instead of the next best alternative. Get this number wrong in one direction and every stock looks cheap. Get it wrong in the other and nothing will ever seem worth buying.

The formal version — WACC — blends after-tax cost of debt with cost of equity, weighted by how much debt and equity the company uses. The cost of debt is fairly stable and estimable from the company's interest expense and tax rate. The cost of equity requires a risk-free rate (usually the 10-year Treasury), a market risk premium (the extra return equities earn over bonds), and a beta (a measure of how much the stock moves relative to the market). Each of those inputs carries its own uncertainty. The equity risk premium in particular has been estimated at anywhere from 3% to 8% depending on the methodology and time period. Feed that range into WACC and you get a range of discount rates that renders the precision theater.

The practical shortcut most value investors use is 10% — roughly in line with the long-run historical return of the US equity market. The logic is direct: if this investment can't beat the market's historical average, you'd have been better off in an index fund. Use 10% as your floor, adjust upward for higher-risk businesses (early-stage, highly leveraged, cyclical with lumpy cash flows), and check your conclusions at both 9% and 11% to see how sensitive the answer is.

The discipline that matters most: pick a rate before you look at the stock, then apply it consistently across everything you evaluate. Investors who adjust the discount rate to make their favorite stock's price work out aren't running a DCF — they're running a confirmation. The discount rate is a prior, not a variable.

Running One on a Real Company

Take a fictional mid-cap industrial — call it Meridian Industrial (MRID). It manufactures aftermarket components for commercial HVAC systems. Revenue is $1.2 billion, growing at 5–7% annually. The business converts about 12.5% of revenue to free cash flow, so trailing FCF is $150 million. Shares outstanding: 100 million. Current stock price: $22.

Step one: define the FCF forecast. HVAC aftermarket is a recurring-revenue business with modest cyclicality — replacement parts, service contracts, predictable demand. You project 7% annual FCF growth for five years, consistent with revenue growth and stable margins. That isn't heroic. Year 1: $160.5 million. Year 5: $210.4 million. If you were more bullish on share gains or margin expansion, you'd use a higher rate. If you were worried about energy-efficiency headwinds reducing replacement cycles, you'd use a lower one. Write down your reasoning before you pick the number.

Step two: set the discount rate. Meridian is stable with moderate leverage. You use 10.5% — slightly above the 10% baseline to reflect mid-cap size and some exposure to residential construction cycles. That choice is judgment. Another investor might use 10.0% or 11%. Those differences will matter.

Step three: estimate terminal value. You assume 2% terminal growth — roughly in line with long-run nominal GDP growth, conservative for a business in a mature but durable sector. Terminal value: $210.4 million × 1.02 ÷ (0.105 − 0.02) = $2,524 million. Discounted back five years at 10.5%, that's $1,527 million. The five years of projected FCF, discounted, sum to about $680 million. Total present value: $2,207 million. Divided by 100 million shares: $22.07 per share.

So at $22, the stock is roughly fairly valued under these assumptions. That's useful information: if you want a margin of safety, you'd need the stock to trade down to $17–18 before your base case pencils out with a 20% discount. It also tells you the market isn't pricing in optimism — it's pricing in exactly your middle scenario.

Now shift two inputs for a bull case: 12% FCF growth for five years, 9% discount rate, 3.5% terminal growth. Intrinsic value jumps to roughly $40 — 80% above the current price. For the bear case: 2% FCF growth, 12% discount rate, 1.5% terminal growth. Intrinsic value falls to about $15 — 32% below where it trades.

The stock is $22. Your base case says fair value. Your bull case says 80% upside. Your bear case says 30% downside. The model didn't tell you what the stock is worth. It told you that the answer hinges almost entirely on two numbers you had to make up — and that the spread between reasonable optimism and reasonable pessimism is not 5% but 60%. That's what the interactive tool above is designed to show you firsthand. Drag the terminal growth slider from 1% to 4% and watch the implied value move. That movement isn't a quirk of the model. That IS the model.

What to Do With the Number You Get

A single-number DCF output is not a buy signal. If your model says $80 and the stock trades at $76, that 5% gap is smaller than the rounding error on your terminal growth assumption. You're not looking at undervaluation. You're looking at noise.

Margin of safety is the discipline that makes DCF actionable. Benjamin Graham's concept is simple: require enough of a discount to your estimated value that being modestly wrong doesn't cost you money. If your base case is $80 and you won't buy until $60 — a 25% discount — you've built a buffer against forecast errors. DCF estimates are imprecise by construction. Margin of safety is how you remain solvent despite that imprecision.

The other powerful use of DCF is reverse engineering. Take the current stock price, plug it in as your output, and solve for the implied terminal growth rate. If a stock at $40 implies 18% annual FCF growth for a decade at a 10% discount rate, you now know exactly what you need to believe to agree with the market. Whether 18% growth is plausible — for this company, in this industry, over this timeframe — is the real investment question. DCF didn't answer it. But it asked it precisely.

Use DCF as a sanity check and a belief-testing machine. Use the range, not the point estimate. Require a meaningful discount to your base case before acting. And treat any model that tells you the stock is worth exactly $39.47 with deep suspicion — that precision is an artifact of the spreadsheet, not a property of reality.