How to Value a Company: The Valuation Framework
Worked example: Hershey — a triangulated valuation
By James Ward
TL;DR: There is no single correct way to value a business. The discipline is to triangulate two or three methods on the same company and read whether they agree. Start by understanding the business, then normalize the cash input to owner earnings rather than reported EPS. Run earnings power value first for a conservative, no-growth floor; run a discounted cash flow for the growth-inclusive case; run a reverse DCF on the current price to see exactly what growth you are being asked to pay for. When all three point to cheap, the signal is strong. When they diverge, that divergence is the thing to investigate before trusting any of them. Finish by applying a margin of safety, sized to how certain the estimate is: durable compounders need a smaller discount, uncertain or cyclical businesses need a larger one. The target of every method is intrinsic value, the discounted value of the cash the business can pay out over its life, and that target is always a range, never a point.
Most people think valuation is a calculation. You feed in the numbers, the model returns a price, and that price tells you what the company is worth. Real valuation works almost the opposite way. The number is the easy part. The judgment is deciding which method fits the business, what inputs are honest, and how much you should distrust your own estimate. A spreadsheet will return a value to the cent for any company you point it at. Whether that value means anything depends entirely on the thinking that went in before the first cell was filled.
The concept in 60 seconds
Intrinsic value is the discounted value of all the cash a business can pay out over its remaining life. It is an economic figure, not an accounting one. It is not book value and it is not the stock price. It is always a range.
No single method captures it reliably across all businesses, because businesses differ. A bank's value lives on its balance sheet; a software company's lives in cash flows it has not earned yet; a holding company's lives in segments that have nothing to do with each other. So the disciplined approach is not to find the one true method. It is to run two or three methods that suit the business and read the agreement between them.
When several independent methods converge on the same answer, your confidence rises. When they scatter, the scatter itself is information: one of your inputs is doing something you have not understood yet. Triangulation does not just produce a number. It produces a number you can trust, or a warning that you should not.
Mental model
Think of valuation like fixing a position at sea before satellite navigation. No single bearing tells you where you are. You take bearings on three separate landmarks and draw three lines. If they cross at one tight point, you know your position with confidence. If they form a wide triangle, you do not average the middle and sail on. You recognize that one of your bearings is wrong, and you find out which before you trust any of them.
Each valuation method is a bearing on intrinsic value taken from a different landmark. Earnings power value reads the business assuming it never grows again. A discounted cash flow reads it assuming a forecast of future cash. A reverse DCF reads the market's own implied expectation embedded in today's price. Three lines, three landmarks. The cross is the answer, and the spread is the risk.
Worked example: Hershey, a triangulated valuation
Take a stable, understandable business such as a dominant confectioner. The work happens in order.
First, understand it. A century-old brand, shelf dominance, pricing power that has tracked inflation, and a product nobody reformulates in a recession. This is the kind of business where the cash flows are predictable enough that a forecast means something. That qualification is itself part of the valuation.
Second, get the input right. You compute normalized owner earnings, the cash the owner could actually extract, rather than headline EPS distorted by a one-time charge or an unusually light capital-spending year.
Third, run earnings power value. Assume zero growth, capitalize the normalized earnings, and you get a conservative floor. Suppose that floor already sits near the market price. That alone is interesting: you are being asked to pay almost nothing for growth in a business that will clearly grow.
Fourth, run a DCF with conservative growth and a deliberate bear case. It returns a value above the EPV floor, as it should for a business that compounds.
Fifth, run a reverse DCF on the current price. It tells you the market is implying a growth rate well below the company's own long history. When the no-growth floor, the forecast value, and the market's implied expectation all line up on the cheap side, the three bearings have crossed at a point. That is a strong signal. Had the DCF said cheap while the reverse DCF said the market was pricing in heroics, you would have stopped and found out why before acting.
Historical pattern
The investors who have compounded capital for decades almost never rely on one model. The pattern that repeats is conservative inputs, more than one method, and a written bear case, applied to businesses simple enough that the forecast is honest. The blow-ups cluster at the opposite corner: a single elaborate DCF, optimistic growth, a terminal value carrying ninety percent of the answer, run on a business whose future nobody could actually forecast. The math was not the failure. The false confidence was.
There is a structural irony worth internalizing early. The businesses that are easiest to value, the predictable compounders where a DCF is most reliable, are exactly the businesses everyone wants, so they rarely trade cheap. The widest discounts to intrinsic value appear in complexity, fear, and uncertainty, which is precisely where valuation is hardest and triangulation matters most.
Decision framework
A workable order of operations:
- Understand the business first. You cannot value what you do not understand. This is research work, done before any number.
- Normalize the input. Use owner earnings, not GAAP EPS and not raw free cash flow.
- Run EPV first. It is fast and gives a conservative, zero-growth floor. Cheap on EPV, proceed.
- Run a DCF for the upside. Conservative assumptions, always with a bear case. See Intrinsic Value and DCF.
- Run a reverse DCF on the price. Understand exactly what growth you are being asked to pay for. See Reverse DCF.
- Triangulate. All three cheap, strong signal. Wide divergence, find the input driving the gap before trusting any of them.
- Apply a margin of safety, scaled to the certainty of the estimate. See Margin of Safety.
- Check the moat. A durable moat, shown by high and stable ROIC, justifies a smaller required discount. No moat demands a larger one.
You can pressure-test the individual steps with the free DCF calculator and reverse DCF calculator.
Common mistakes
- Hunting for the one true method. There is none. The method is fitting the right two or three to the business and reading the agreement.
- Confusing precision with accuracy. A value to the cent is not more accurate, only more confident about the same uncertain inputs. Present a range.
- Treating growth as a separate bucket. Growth is a component of value, not a bonus on top. A business reinvesting at high ROIC is worth more today; the only question is whether the price reflects it.
- Averaging a wide triangle. If your methods disagree sharply, the answer is not the midpoint. It is to investigate the disagreement.
- Feeding GAAP EPS into the model. Garbage in compounds. Normalize first.
How VI Stack uses this
VI Stack runs this exact order of operations inside Gate 3 and Gate 4 of the research process. The system normalizes owner earnings, runs the conservative floor before the growth case, inverts the price into an implied expectation, and forces an explicit bear case rather than a single point estimate. The output is never a single number presented as truth. It is a range, the methods behind it, and the margin of safety the current price does or does not offer. The discipline is built into the sequence, so the judgment happens in the right order and the false confidence has nowhere to hide.
What's next
The methods in this framework each have their own article. Start with Intrinsic Value and DCF for the standard method on predictable businesses, then Reverse DCF to read what the market already implies, and Margin of Safety for the discount that protects you when, not if, your estimate is wrong.
FAQ
What is the best way to value a company?
There is no single best method. The disciplined approach is to triangulate two or three methods suited to the business and read whether they agree. For a predictable business, that usually means earnings power value for a conservative floor, a discounted cash flow for the growth case, and a reverse DCF to check what the current price already implies. When all three point to cheap, the signal is strong; when they diverge, the divergence tells you an input needs investigating before you trust any single answer.
What is intrinsic value?
Intrinsic value is the discounted value of all the cash a business can pay out to its owners over its remaining life. It is an economic estimate, not an accounting figure, so it is neither book value nor the current stock price. Critically, it is always a range rather than a precise point, because it depends on assumptions about future cash, growth, and the discount rate that no one can know exactly.
Should I use one valuation method or several?
Several. Any single method rests on assumptions that can be wrong, and a single number gives false confidence. Running two or three independent methods on the same business and comparing them either confirms the value from different angles or exposes a flawed input. The agreement between methods is the real output, not the number from any one of them.
Why normalize earnings before valuing a company?
Because reported earnings in any given year can be distorted by one-time charges, unusually high or low capital spending, or accounting choices that do not reflect the business's true earning power. Valuing a single distorted year values the year, not the business. Normalizing to owner earnings, the sustainable cash the owner could actually extract, gives every valuation method an honest input to work from.
How big should the margin of safety be?
It scales with how certain your estimate is. A durable, predictable business with a tight range of outcomes might justify buying at a 25 to 30 percent discount to a conservative intrinsic value. A cyclical or harder-to-forecast business demands a deeper discount before the odds are in your favor. The margin is a judgment about your own uncertainty, not a fixed percentage applied to every company.
vistack.io