How to Value a Stock with Discounted Cash Flow

What DCF actually does

The standard explanation of DCF frames it as a calculation: project the company's future cash flows, discount them back to today, add a terminal value, divide by shares outstanding, and you have intrinsic value. That framing is technically correct. It is also the source of most DCF mistakes, because it makes the model sound like it produces a fact.

DCF produces a number that is exactly as reliable as the three sets of estimates you fed it. None of those estimates can be derived from a financial statement. Two analysts with identical access to the same 10-K, making defensible but different judgment calls about future growth and risk, will reach intrinsic value estimates that differ by 20–40%. That's not a failure of the tool. That is the tool.

The useful reframe: DCF is a structured way to make your assumptions explicit so you can test whether they hold up under pressure. Run it forward to estimate value. Run it backward from the current stock price to reverse-engineer what the market is already pricing in. The model gives you the question. You supply the judgment about whether the answer makes sense.

The four numbers you need, in plain English

Every DCF rests on four inputs. They matter in sequence — each one sets up the next. To keep the mechanics concrete, run these through a single illustrative example: Meridian Industrial (MRID), a fictional mid-cap that makes aftermarket components for commercial HVAC systems. Revenue is $1.2 billion. 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.

Free cash flow is what you're forecasting — what the business generates in cash after paying for operations and reinvestment. Take operating cash flow from the cash flow statement and subtract capital expenditures. Both appear on every 10-K and 10-Q. For Meridian, that's $150 million. Some analysts adjust for stock-based compensation (non-cash but real dilution) or normalize for unusual working capital swings. Those adjustments can be legitimate — just document the reasoning and apply it consistently. The goal is a number that represents what the business actually earns in cash before any financing decisions.

The near-term growth rate is your forecast for how fast FCF grows over the next five to ten years. Meridian's market is driven by replacement cycles and commercial construction activity — recurring demand, modest cyclicality. You project 7% annual growth, consistent with revenue growth and stable margins. Write down why you chose that number before moving on. If you can't explain it in one sentence, the choice is not ready.

The discount rate converts future dollars into today's dollars. At 10%, a dollar of FCF arriving five years from now is worth about $0.62 today. The rate reflects the return you require to take on this investment's specific risks — how predictable the cash flows are, how levered the balance sheet is, how exposed the model is to conditions outside the business's control. For Meridian, 10.5% feels right — slightly above the 10% baseline to reflect mid-cap size and modest exposure to commercial construction cycles.

Terminal value is the fourth input and the largest single driver of the output. After your five- or ten-year forecast, terminal value captures everything that happens after that — years 6 through infinity, expressed as a single number. The formula is FCF₅ × (1 + g) ÷ (r − g), where g is the terminal growth rate and r is the discount rate. At 2% terminal growth and 10.5% discount, Meridian's terminal value works out to approximately $2,524 million. Discounted back five years, it's worth about $1,527 million today. Add the discounted sum of years one through five (~$680 million), and total intrinsic equity value is $2,207 million — or $22.07 per share. Exactly in line with where the stock trades, under these assumptions.

Terminal value: the number that swallows the model

That 69% figure above is not a quirk of Meridian's profile. It is how DCF math works. For a typical company with stable cash flows, terminal value represents 60–75% of the total output. For a fast-growing software business with minimal near-term FCF, it can exceed 85%. You can spend three days arguing over the revenue forecast and still get the valuation wrong because you rounded the terminal growth rate.

The math behind the sensitivity is direct. The terminal value formula divides by (r − g). At a 10% discount rate and 2% terminal growth, that denominator is 8%. Raise terminal growth by a single percentage point to 3%, and the denominator drops to 7% — a 12.5% smaller number, which means terminal value grows by more than 12.5%. On Meridian, that one-point move — from 2% to 3% — pushes intrinsic value from $22 to roughly $27. That's a 23% increase from a single rounding decision on a number no spreadsheet can derive from first principles.

This is the core finding that most DCF tutorials bury: two analysts running honest models on the same company, using the same trailing FCF and the same near-term projections, can produce intrinsic value estimates 23–40% apart purely because they chose different terminal growth assumptions. Neither made an error. They made different bets about what happens in year 10 and beyond. When you hear two Wall Street analysts arguing about the same stock, reaching opposite conclusions from the same earnings report — this is almost always the source. Not a dispute about Q3 results. A dispute about a number in a formula's denominator.

One sanity check on terminal growth: it cannot sustainably exceed the long-run nominal growth rate of the economy. The US economy has grown roughly 5–6% nominally over recent decades. A company growing faster than the whole economy forever eventually becomes the whole economy. For most businesses in established sectors, 2–3% is the honest answer. If a DCF only pencils out at a 4% or higher terminal growth rate, that's something to interrogate — not a bullish datapoint.

How to pick a discount rate

The formal answer is WACC — a blend of after-tax cost of debt and cost of equity, weighted by capital structure. Cost of debt is estimable from interest expense and tax rate. Cost of equity requires a risk-free rate, a market risk premium, and a beta. The market risk premium has been estimated at anywhere from 3% to 8% depending on methodology and time period. Feed that range into WACC and the precision of the formula becomes an artifact, not a signal.

The practical shortcut most value investors actually use:

• Large, established US company with predictable cash flows: 8–10%. The 10% floor reflects the long-run historical return of US equities — your opportunity cost.
• Mid-cap or businesses with meaningful cyclicality: 10–12%. Size and volatility warrant a higher required return.
• Smaller companies, highly levered, or sectors undergoing structural change: 12–15%. Predictability is genuinely lower; the discount rate should reflect that.
• Pre-revenue growth stocks: DCF is probably the wrong primary tool (next section). If you run one anyway, 15%+ is the starting point.

The discipline that matters more than the specific number: set the discount rate before you look at the stock. Investors who adjust the rate until the model produces the answer they wanted are not running a DCF. They are running a confirmation. The discount rate is a prior. It is not a variable you tune to get the result you like. If someone hands you a DCF on a cyclical industrial at a 6% discount rate, ask why before engaging with the price target.

Where DCF breaks down

DCF is a cash flow machine. It works well on businesses where cash flows are positive, predictable, and growing from a stable base. When those conditions are absent, the model produces numbers that feel precise but are speculation formatted as arithmetic.

• Negative free cash flow with no clear path to breakeven. If you cannot project when FCF turns positive, projecting it forward produces fiction. Early-stage companies, most biotechs pre-approval, and genuine turnarounds in process fall here. Use EV/Revenue comps or a probability-weighted scenario model instead.
• Pre-revenue businesses. A DCF on a company with $0 in current cash generation is a story dressed up as math. The terminal value will represent 95%+ of the output and will be entirely determined by an assumption you cannot defend with data.
• Highly cyclical businesses at peak cycle. Running a DCF on a semiconductor company using last year's record FCF as the base and projecting it forward is a reliable path to overpaying. The base number entering your forecast needs to be normalized to midcycle earnings — not the recent peak. For cyclicals, DCF works, but only from a through-cycle estimate.
• Businesses in structural decline. The Gordon Growth Model requires a positive terminal growth rate. A business where the industry is contracting — certain legacy media formats, shrinking commodity producers — does not fit. A negative g blows up the denominator in a different direction. The model was not designed for shrink.

When DCF doesn't fit, comparable company analysis against peers or a sum-of-the-parts gives you more honest anchoring. The model doesn't fail because it's poorly designed. It fails when you apply it to a business it wasn't designed for.

Try it: how sensitive is the price to your assumptions?

Reading about assumption sensitivity is one thing. Dragging the terminal growth rate from 1.5% to 3.5% and watching the implied per-share value jump from $16 to $29 on the same starting cash flows is a different kind of understanding. The tool below uses Meridian's base case — $150M trailing FCF, 100M shares, 7% near-term growth. Move the discount rate or terminal growth rate and the implied value responds immediately. The percentage of total value sitting in terminal value updates in the corner.

Find a combination that makes the stock look cheap. Then notice which assumption you had to use to get there. That assumption is your thesis. It's the thing that has to be right for the investment to work.

How to read a DCF output like an analyst

A professional equity analyst does not hand over a single DCF price target and say "this is what the stock is worth." They run three cases — bear, base, bull — and report a range. The base case is the most likely outcome. The bull case tests what happens if two or three key assumptions come in better than expected. The bear case asks how bad things get if they don't.

For Meridian at $22: the base case gives $22.07 — roughly in line with market price. A bull case at 12% FCF growth, 9% discount, 3.5% terminal growth gives roughly $40. A bear case at 2% FCF growth, 12% discount, 1.5% terminal growth gives roughly $15. The stock trades at $22. The model didn't tell you what to do. It told you the spread: 80% upside if the bull case materializes, 30% downside if the bear case does. The investment decision is a judgment about which outcome is more likely, not a read on which number the formula produced.

The next step a real analyst takes: check the output against comps. If Meridian's DCF implies fair value at $22 and every comparable public company trades at 18–20× FCF (implying $27–30), the discrepancy matters. Either the DCF assumptions are too conservative, or the comps are overvalued, or there's something specific to Meridian — a litigation overhang, a margin compression cycle, a CEO transition — that justifies a discount. Identify the source of the gap before deciding. The comparable company analysis is not a replacement for DCF; it's a sanity check on whether your assumptions are in the same neighborhood as what the market charges for similar businesses.

The most powerful use of the model is running it in reverse. Take the current stock price, set it as the output, and solve for the implied terminal growth rate. If Meridian at $22 implies 2% perpetual growth and you believe 2.5% is realistic for a business with an installed base of 40,000 commercial HVAC systems, you've found the gap. Your edge isn't "my DCF says $27." Your edge is "the market is pricing in 1.5% terminal growth for a business with structural replacement demand, and I think that's wrong — and here is specifically why." That thesis you can defend. A price target you can argue about forever.

One last discipline: require a margin of safety before acting. A DCF estimate is not a measurement — it is a range with a most-likely midpoint. If your base case is $27 and the stock trades at $26, the difference is inside the model's error bars. Fifteen to twenty-five percent is a common threshold. It means that if you're modestly wrong on the assumptions, you don't lose money. It also means the stock has to be genuinely cheap by your own analysis — not just modestly below a number you derived from assumptions you chose yourself.