Why are a bank and a software company valued so differently?

A software firm might trade at 50 times earnings while a bank trades at 10. It's tempting to call the bank "cheap" and the software firm "expensive" — but that comparison is a beginner's trap. The gap isn't a mispricing; it's the market pricing two genuinely different futures. Understanding why multiples differ across sectors is the heart of this course.

Multiples encode expectations

A price multiple is a compressed statement about a company's future. Three fundamentals set the level a whole sector deserves:

• Growth. Faster-growing profits are worth more per current dollar of earnings — so high-growth sectors (much of Technology) carry high multiples. You're paying today for earnings expected to be far larger tomorrow. (This is your intrinsic-value discounting, viewed from the multiple side.)
• Stability / risk. Steadier, more predictable earnings deserve a higher multiple than lumpy, cyclical ones — certainty is worth paying for. A cyclical carmaker's earnings could halve in a recession, so the market won't pay a rich multiple for them.
• Capital intensity and returns. A business that grows without swallowing cash (asset-light software) is worth more per dollar of profit than one that must constantly reinvest in factories or hold huge capital (a bank, a utility).

Stack these up and the "correct" multiple for a sector falls out. Software: high growth, asset-light, high returns → high multiples. Banks: modest growth, heavily regulated, capital-intensive, cyclical credit risk → low multiples. The gap is information, not a bargain.

Different sectors, different yardsticks

It goes further: sectors are often valued on different multiples entirely, because different fundamentals matter.

• Banks are typically read on P/B (price-to-book), because a bank essentially is a pile of financial assets and liabilities — book value is the natural anchor, and return on equity tells you how well they work it.
• Software / growth firms are often read on P/S or forward earnings, because current profit understates a business deliberately spending to grow.
• Capital-heavy sectors (telecom, industrials) lean on EV/EBITDA, which sees past their big debt loads and depreciation.
• REITs get their own measure (funds from operations) because standard earnings mangle property accounting.

So comparing a bank's P/E to a software firm's P/E isn't just apples-to-oranges — it's often using the wrong ruler for at least one of them.

A worked example

NimbleSoft trades at a P/E of 45; SolidBank at a P/E of 10. The naïve read: sell the "expensive" software, buy the "cheap" bank. The real read: NimbleSoft's profits are expected to grow ~30% a year, arrive with high margins, and need almost no capital — so a high multiple is rational. SolidBank grows slowly, is capital-bound and cyclical — so a low multiple is rational too. Both can be fairly priced at the same time. The multiples aren't ranking quality; they're describing two different futures. To find a real bargain, you compare NimbleSoft to other software and SolidBank to other banks — never across the aisle.

In the data

Data services publish the same seven multiples for every stock, whether or not each one means anything for that business: trailing and forward P/E, price-to-sales, price-to-book, enterprise value, EV/sales and EV/EBITDA. Here is the set for a bank:

Price-to-book comes back as a normal figure, but EV/EBITDA reads 0. A bank has no meaningful EBITDA, and instead of leaving the multiple out, the table prints a zero. A ranking sorted from lowest EV/EBITDA up therefore fills its top with names that never earned the multiple: Rivian at −1.955 for a negative EBITDA, then every bank at 0, and only then a company whose figure means something (read 29 September 2026).

Try it now

An asset-light technology company, on exactly the same seven measures as the bank above:

1. Note both trailing P/E figures, then both price-to-book figures. The two rankings do not agree, and neither is the "right" one — book value anchors the bank because a bank essentially is a pile of financial assets; for the other company it barely describes anything.
2. Now read the last row of the bank's table. EV/EBITDA reads 0: a value, sitting in the table, with nothing beside it to say the multiple does not apply here. Say out loud what a zero in that position would mean if you believed it, and you have the reason the section above calls this the dangerous case rather than the missing one.
3. Confirm the trap yourself. Five names across three sectors, on the same multiple:

Sort them from the lowest EV/EBITDA up and look at what floats to the top. Count how many names come before the first one whose figure means something, and say why each of them is there. 4. Say the course's central sentence out loud: a multiple is an expectation about growth, risk, and capital — so different sectors deserve different multiples, and comparing across sectors compares different futures.