Every valuation multiple hides a discounted cash flow

A price divided by a fundamental

Relative valuation prices an asset by what the market pays for comparable assets.¹ Prices cannot be compared directly. A $224 share and an $87 share tell you nothing about which business is dearer, because a share is an arbitrary slice of a company. So you divide the price by some fundamental the business produces, earnings or book value or revenue, and compare the ratios instead. That ratio is a multiple.¹

The most common one is the price-to-earnings ratio, or P/E: the market price of a share divided by the earnings per share behind it.² Take NVIDIA. For its fiscal year ended January 25, 2026, it reported diluted earnings of $4.90 a share.³ At its close of $223.96 on August 7, 2026, that is a price of 45.7 times those earnings, call it 46.⁴ The same market data service also lists NVIDIA at 34 times its trailing twelve months of earnings and 22 times the earnings analysts expect next year.⁴

One price, three multiples, and they disagree by a factor of two.⁴ Nothing changed about the company between them. Only the denominator moved, from the year that ended in January, to the most recent twelve months, to the year ahead, and NVIDIA's earnings are climbing fast enough that the choice nearly doubles the number. All three are honestly called NVIDIA's P/E. This is the first discipline of multiples: name the denominator every time, because a multiple without one is ambiguous.

There is a second discipline. The numerator and the denominator must belong to the same owners.¹ A share price is the value of equity, and earnings per share belong to equity holders, so the P/E is an equity multiple, consistent on both sides. To value the whole business rather than just its equity, you use enterprise value over EBITDA. Enterprise value is the market value of equity plus debt minus cash. EBITDA is earnings before interest, taxes, depreciation, and amortization, the profit available to everyone who financed the firm.² Both belong to all the capital, debt and equity together. Divide a value that belongs to shareholders by earnings that belong to the whole firm and the ratio means nothing at all.

The multiple a business deserves

The market already quotes a multiple, so why derive one from a model? Because the quoted number is an answer with its reasoning stripped off. Deriving a multiple shows what it would have to be if the business is worth the cash it will hand its owners, and that is what lets you say whether the quoted number makes sense. The derivation is short, and it uses only what the earlier lessons already built.

Start where the discounted cash flow lesson ended. A stable business whose payouts grow at a rate g forever is worth, in equity terms, next year's payout divided by the cost of equity r minus that growth. The payout is the earnings times the share of them paid out, the payout ratio. Divide both sides by next year's earnings and the price becomes a forward P/E.²

One more substitution ties the payout ratio to how the business runs. A company grows by keeping some earnings and reinvesting them, and the growth it can sustain is the share it keeps times the return it earns on equity, its ROE. So the growth rate fixes the payout: whatever a business does not need to retain to fund its growth, it can pay out, which is 1 - g/\mathrm{ROE}.² Put that in place of the payout ratio and the multiple stands on three fundamentals alone.

Read the formula and the whole logic of pricing falls out of it. The multiple rises with growth. It rises with return on equity, because a business that earns more on each retained dollar can pay out more and still grow. And it falls as risk rises, because a larger r in the denominator shrinks the whole fraction.² Growth, returns, and risk are exactly the inputs a discounted cash flow model turns. They have not disappeared into the multiple. They are folded inside it.¹

Enterprise value over EBITDA has the same shape, with the cost of capital in place of the cost of equity and the reinvestment a firm needs subtracted in the numerator.² The lesson is the same either way. A multiple is a discounted cash flow compressed into a single number.

Three companies, three multiples

Take three businesses whose growth alone spreads them far apart. NVIDIA grew revenue 65% in its last fiscal year. The Coca-Cola Company grew about 2%, and Verizon about 2.5%.³⁵⁶

Start with Coca-Cola, because it is the case the formula explains most cleanly. It grew revenue under 2% as reported, about 5% organically, and yet it holds a trailing multiple near 26.⁵⁴ A reader trained only to think low growth means a low multiple should find that strange. The formula does not. Coca-Cola earned $13.1 billion on $32.2 billion of equity in 2025, a return on equity near 41%.⁵ Suppose its earnings grow 4% over the long run and its shareholders require 7%, a low rate for a business this steady. The justified forward P/E is then (1 - 0.04/0.41)/(0.07 - 0.04), which works out near 30, in the neighborhood of what the market quotes.² The growth rate and the cost of equity there are my own estimates, not figures any filing owns, and a point either way moves the answer. What holds regardless is the mechanism. A business that earns 41 cents on the equity dollar barely needs to retain anything to fund 4% growth. So it returns almost all of its earnings to owners, and each earnings dollar is worth a great deal.

NVIDIA does not fit the same plug, and it should not. The formula describes a business in stable growth, and a company that grew 65% in a year is nowhere near stable.³ You would value it in two stages, a stretch of high growth and then a settled tail, which is why its full derivation is longer.² The two-stage story is already visible in the table. NVIDIA trades at 34 times its trailing earnings but only 22 times next year's.⁴ The market is pricing earnings it expects to be much larger a year out. The high multiple is the growth, seen through whichever denominator you pick.

When a low multiple is not cheap

Verizon is the company on the table that looks like a bargain. It trades near 12 times earnings, less than half of Coca-Cola's multiple and a third of NVIDIA's on trailing earnings.⁴ Cheap compared to what? Run it through the formula. Verizon earned about 16% on its equity in 2025.⁶ Suppose 2.5% long-run growth and a cost of equity of 8.5%, higher than Coca-Cola's because Verizon carries far more debt and more exposure to interest rates. The justified forward P/E is (1 - 0.025/0.16)/(0.085 - 0.025), which is about 14, close to what the market pays.² A lower return on equity and a higher discount rate pull the deserved multiple down to roughly where the observed one sits.

The fundamentals are harder than slow, too. Verizon's earnings per share fell in 2025, to $4.06 from $4.15, and net income attributable to the company slipped to $17.2 billion from $17.5 billion.⁶ It carries about $158 billion of debt and spent $17 billion on capital in the year to hold its position.⁶

The formula gives the multiple a business deserves. It does not prove the market's quoted multiple equals that number. Relative valuation prices a business off its comparables, so it inherits whatever mistake is priced into those comparables; if a whole sector is dear, every multiple in it looks reasonable against the others.¹ The safe claim is the narrow one. Different fundamentals justify different multiples. A low multiple is a hypothesis about a company's growth, returns, and risk, waiting to be tested against those numbers. It is not a verdict that the stock is cheap. To compare two multiples honestly is to compare the two discounted cash flows hidden inside them.¹