Value Investing
Estimating Intrinsic Value in Practice
Run the discounted cash flow model end to end on a real company's published figures, then learn why the number it gives you is not a reason to buy.
Putting the pieces together
Last lesson you met every input the discounted cash flow (DCF) model needs — a way of estimating what a business is worth by projecting the cash it will produce and converting that future cash into today's money. You learned where growth rates come from, why a terminal value stands in for everything past your forecast, why a discount rate shrinks distant money more than near money, and how the capital asset pricing model (CAPM) gives you a defensible rate to discount with. You also learned which direction each input pushes: faster growth or a higher terminal value raises your estimate, a higher discount rate lowers it.
You have the parts. This lesson assembles them.
There's no spreadsheet to download here, and that's deliberate. A calculator that hides the arithmetic teaches you to trust an output you can't check. Instead we're going to run the whole model once, by hand, on a real company using figures it actually published. Follow it with a piece of paper or any spreadsheet you like. Once you've done it once, you can do it for any company — and more importantly, you'll know which assumption to blame when the answer looks strange.
There's nothing to download here, and that's on purpose. A calculator that hides the arithmetic teaches you to trust an answer you have no way to check. So instead we're going to run the whole thing once, by hand, on a real company, using figures it actually published. Follow along with paper or any spreadsheet you like. Once you've done it a single time you can do it for any company — and, more useful than that, you'll know which guess to go blame when an answer comes out looking strange.
There's nothing here to download, and that's on purpose. A calculator that hides the arithmetic just teaches you to trust an answer you can't check.
So we're going to do the whole thing once, by hand, using numbers a real company actually published. Follow along on paper if you like.
Do it once and two things happen. You can do it for any company after that. And — the more useful one — when an answer comes out looking strange, you'll know exactly which guess to go blame.
Video coming soon
This lesson explains the idea in full without it.
The six steps
Every DCF estimate, no matter how elaborate, is these six moves in this order.
| Step | What you do |
|---|---|
| 1 | Start with a current measure of the company's cash flow — commonly earnings per share (EPS) for the last complete fiscal year, or over the trailing twelve months (TTM), meaning the four most recent quarters the company has actually reported. |
| 2 | Project that figure forward five years using a growth rate you can justify, such as the company's own history or its published guidance. |
| 3 | Estimate a terminal value at the end of year five by applying an exit multiple — a price multiple, often the industry average price-to-earnings (P/E) ratio, applied to your last projected year. |
| 4 | Pick your discount rate. CAPM gives you one: the return an investor should require for carrying this company's market risk. |
| 5 | Discount each of the five projected years, and the terminal value, back to today. |
| 6 | Add the six discounted amounts together. That sum is your estimated intrinsic value per share. |
Steps 1 through 4 are judgment. Steps 5 and 6 are arithmetic. Almost everything that goes wrong with a DCF goes wrong in the first four.
We're about to prove that, rather than just assert it.
Step 1 has a fork in it
Textbooks start a DCF by saying "take EPS." Real companies hand you two of them.
For fiscal 2025 — Campbell's fiscal year ended August 3, 2025, which is why it doesn't line up with the calendar — the company reported:
| Measure | FY2025 diluted EPS |
|---|---|
| GAAP — the figure required by standardized accounting rules | $2.01 |
| Adjusted — the company's own preferred figure, with certain items removed | $2.97 |
Those describe the same fiscal year of the same business, and they are 48% apart.
The gap is made of items Campbell's excluded from its adjusted figure. The two largest: $0.44 a share of impairment charges — writing down the carrying value of assets and brands the company had booked at more than they now appear to be worth — and $0.32 a share of costs from its cost-savings program. Smaller amounts came from accelerated amortization, divestitures, litigation, and pension accounting.
Neither number is a lie. Management's argument for the adjusted figure is that impairments and restructuring costs obscure how the business performs in an ordinary year. The argument against it is that a company choosing which of its own expenses count is not a neutral party, and "unusual" items that appear year after year are not unusual — they're the cost of running that particular business.
You have to pick one to start with. Nobody picks it for you.
Neither number is a lie. The company's argument for its own adjusted figure is that write-downs and reorganization costs get in the way of seeing how the business does in an ordinary year. The argument against it is that a company deciding which of its own expenses count is not a neutral referee — and "unusual" costs that turn up year after year are not unusual. They're what it costs to run that particular business.
You have to pick one of the two to start from. Nobody picks it for you.
Neither number is a lie.
The company's argument for its own figure goes like this: some of what we spent that year was unusual — one-off things that get in the way of seeing how we do in a normal year.
The argument against it goes like this: a company deciding for itself which of its own costs count is not exactly a neutral referee. And "unusual" costs that show up year after year after year aren't unusual at all. They're just what it costs to run that particular business.
You have to pick one of the two to start from. Nobody picks it for you.
We'll run the model on the GAAP $2.01, because it's the figure the accounting rules produce rather than the one the company would prefer you use. Then, at the end, we'll run it again on the adjusted number and see what that single choice was worth.
Running the six steps
Step 1 — the starting cash flow. FY2025 GAAP diluted EPS: $2.01. On roughly 300 million diluted shares against net sales of $10.25 billion. One wrinkle you'd only catch by reading the release: fiscal 2025 ran 53 weeks instead of the usual 52, and the company estimates that extra week added about $0.06 to its adjusted EPS. Reporting periods aren't all the same length, and that never appears on the summary page you'd get from a stock screener.
Step 2 — project five years. Here judgment starts. Campbell's has guided fiscal 2026 adjusted EPS to $2.15–$2.25, which the company itself describes as a decline of 23% to 26% from fiscal 2025. A business its own management expects to earn less next year is not one you project double-digit growth from. We'll assume earnings hold flat at $2.01 for five years — an assumption that is already more generous than the near-term guidance, and one you should be prepared to defend.
| Year | Projected EPS |
|---|---|
| 1 | $2.01 |
| 2 | $2.01 |
| 3 | $2.01 |
| 4 | $2.01 |
| 5 | $2.01 |
Step 3 — the terminal value. We'll use an exit multiple of 11. Multiply it by year five's EPS:
11 × $2.01 = $22.11
That $22.11 is a rough stand-in for what someone might pay for a share five years from now, given everything the business is expected to earn after that. Where does 11 come from? It's roughly where large, slow-growing packaged-food companies have traded — which is to say it's a judgment call dressed as a fact, and step 3 is where most DCF answers are quietly decided.
Step 4 — the discount rate. We'll use 9% — a defensible rate for a large, stable, dividend-paying consumer-staples business, which is less volatile than the market as a whole. That's what an investor should want per year for taking this risk, and it's the rate we'll use to drag those future dollars back to today.
Step 5 — discount everything. Divide each future amount by 1.09 raised to the number of years away it is. Year three gets divided by 1.09 × 1.09 × 1.09, and so on.
| What | Amount | Years away | Divided by | Value today |
|---|---|---|---|---|
| Year 1 EPS | $2.01 | 1 | 1.090 | $1.84 |
| Year 2 EPS | $2.01 | 2 | 1.188 | $1.69 |
| Year 3 EPS | $2.01 | 3 | 1.295 | $1.55 |
| Year 4 EPS | $2.01 | 4 | 1.412 | $1.42 |
| Year 5 EPS | $2.01 | 5 | 1.539 | $1.31 |
| Terminal value | $22.11 | 5 | 1.539 | $14.37 |
Step 6 — add them up. $1.84 + $1.69 + $1.55 + $1.42 + $1.31 + $14.37 = $22.18.
Your estimated intrinsic value is about $22.18 a share. That's the whole model.
Figure
Look at that last column before you move on. The five projected years contribute $7.81 between them. The terminal value contributes $14.37 — nearly two-thirds of the estimate. That is normal, and it is the most uncomfortable fact about the DCF model: most of your answer comes from a single number built on an exit multiple you picked.
What each assumption was worth
Now go back and change one input at a time, leaving everything else alone. Same company, same reported earnings, same five years.
| Change one input | Estimate | vs. the price |
|---|---|---|
| Baseline — GAAP $2.01, 0% growth, 11× exit, 9% discount | $22.18 | 1% above |
| Start from adjusted EPS ($2.97) instead of GAAP | about $32.80 | 49% above |
| Exit multiple 15× instead of 11× | about $27.40 | 25% above |
| Growth 3% a year instead of 0% | about $25.20 | 14% above |
| Discount rate 11% instead of 9% | about $20.60 | 7% below |
| Earnings shrink 3% a year instead of holding flat | about $19.50 | 11% below |
| Exit multiple 8× instead of 11× | about $18.30 | 17% below |
One company. One set of reported results. Estimates from $18.30 to $32.80 — a spread of 80% — and every row is defensible enough to say out loud.
Look at the second row hardest. That's not an opinion about growth or risk. That's the very first number you write down, before any arithmetic happens at all, and choosing the figure management prefers over the one the accounting rules require adds nearly eleven dollars a share. Steps 1 through 4 are judgment. This is what that sentence costs.
One more comparison, because you've seen these numbers before. Last lesson's fictional UVWXYZ started from $2.00 of EPS — one penny less than Campbell's — and came out worth $40.00. Run Campbell's real $2.01 through that lesson's assumptions (8% growth, a 15× exit, an 8% discount) and you get $40.20. Run UVWXYZ's $2.00 through this lesson's — flat earnings, 11×, 9% — and you get about $22. A penny of earnings separates the two companies; eighteen dollars of estimate separates the two sets of assumptions. The model never noticed which company was real.
Your estimate is not a price
Here is the part that matters more than the arithmetic.
You now have $22.18. On July 31, 2026, the stock closed at $21.98.
After all that work — four judgment calls, six steps, a table of discount factors — the model says the shares are worth about what they cost. Twenty cents apart. Not a bargain, not a warning, just a number that has landed on top of the number it was supposed to test.
That is the most common honest outcome of a DCF, and it is the one no textbook prints. Most of the time the market has already thought about the company roughly as hard as you have. When your estimate and the price disagree by a lot, the honest first thought is not "the market is wrong." It's "one of us is, and I built mine out of assumptions this morning."
After all that work — four judgment calls, six steps, a whole table of arithmetic — the model says the shares are worth about what they cost. Twenty cents apart. Not a bargain, not a warning. Just a number that came to rest on top of the number it was supposed to be testing.
That is the most common honest outcome of this exercise, and it's the one no textbook prints. Most of the time, other people have already thought about the company roughly as hard as you have. So when your estimate and the price disagree by a lot, the honest first thought is not "the market is wrong." It's "one of us is, and I built mine out of assumptions I picked this morning."
After all that work — four judgment calls, six steps, a whole table of arithmetic — the answer landed almost exactly on top of the price. Not a bargain. Not a warning. Just a number that came to rest on the very number it was supposed to test.
That is the most common honest ending, and it's the one nobody puts in a book.
Most of the time, other people have already thought about the company about as hard as you have. So when your answer and the price disagree by a lot, the honest first thought isn't "everybody else is wrong." It's "one of us is — and I built mine out of guesses I made this morning."
What turns an estimate into a decision is the margin of safety — the gap between your estimated intrinsic value and the price you'd actually pay, expressed as a percentage of your estimate. It exists because you will be wrong. Not might be. Will be, regularly, about growth rates and multiples and how the next five years go. The margin of safety is the room you leave for that.
Here:
($22.18 − $21.98) ÷ $22.18 = 0.9%
Under one percent. If your plan requires 25% — the next lessons cover how to set that threshold and where it goes in your plan — this is not close. Against a $22.18 estimate you'd need the price down at $16.64. The answer isn't "buy" or "avoid." It's "this exercise didn't produce a reason to act," which is a complete and useful result.
Run the loss case too. Suppose you buy at $21.98 and the flat-earnings assumption turns out generous — earnings shrink 3% a year instead of holding steady, which is milder than what the company's own FY2026 guidance describes. Rerun the six steps and the estimate lands near $19.50, below the price you paid. The company you thought was fairly valued was expensive, and the market saw something you didn't.
A margin of safety wouldn't have made that forecast right. It would have meant you only bought at a price where being that wrong still left you somewhere to stand.
What to carry out of here
The six steps are a discipline, not an oracle. They force you to state what you believe about a company's growth, its risk, and what it'll be worth when you're done — in numbers, where you can be checked. That's worth doing even when the answer is "I don't know enough to fill in step two," because that's an answer too, and it's the one that saves you money.
And notice what the exercise actually produced. Not a verdict. A range from $18.30 to $32.80, a market price sitting inside it, and a clear account of which of your own assumptions moved the answer most. That range is the finding. Anyone who runs this model and reports a single number to three decimal places has thrown away the only honest part of it.
Nothing here tells you what to buy. It tells you how to price your own beliefs, and how much room to leave for the strong possibility that they're wrong.
Key takeaways
- A DCF estimate is six steps: start with current cash flow, project it five years, estimate a terminal value with an exit multiple, pick a discount rate, discount everything to today, and add it up.
- Step 1 is already a judgment call. Companies report both a GAAP earnings figure and their own adjusted one, and they can differ enormously. Read the reconciliation and decide which you're using before you start.
- Steps 1 through 4 are judgment; only steps 5 and 6 are math. Errors live in the judgment. Choose your assumptions before you look at the price.
- The terminal value usually dominates the estimate, so the exit multiple you pick moves the answer more than almost anything else. Small assumption changes produce large valuation changes.
- Your estimate is not a price, and landing near the market price is the most common honest result. A disagreement with the market is not a buy signal by itself.
- The margin of safety decides — the gap between your estimate and the price, sized to absorb the fact that you will sometimes be badly wrong. A price merely below your estimate is not enough.
- The method is six steps: start with what the company earns now, guess it forward five years, guess what the whole thing is worth at the end, decide how much to shrink future money, shrink it, and add everything up.
- The very first step is already a judgment call. A real company publishes two different profit numbers for the same year, and they can be far apart. You have to choose one before any arithmetic happens.
- Only the last two steps are actually math. The first four are guesses. That's where things go wrong — so make your guesses before you look at the price, not after.
- Most of the final answer comes from one guess about what the business is worth at the very end. That single choice moves the answer more than almost anything else you do.
- Your answer is not a price. Landing right on top of the price is the most common honest result, and disagreeing with everyone else is not by itself a reason to buy.
- What decides is the gap between your answer and what you'd pay — big enough to survive being badly wrong. A price a little under your number is not enough.
Check your understanding
Question 1 of 5