ReportsMonitoringPortfolioMethodologyPricingBlogAboutStart for free
Back to blog

Analysis

What Is a PE Ratio, and Where Each Multiple Breaks

What is a PE ratio, what price to sales, price to book and price to operating cash add, the point where each of them breaks, and how to set your own band.

14 min read
See a sample report

This is a method, not a recommendation. Nothing here, or anywhere else on Taufolio, is investment advice. Treat every example as a starting point for your own research.

A PE ratio does not tell you what a company costs. It tells you how many years of today's profit you are paying for one share, on the assumption that the profit never changes again. It always changes, and everything difficult about this number follows from that one sentence. Like every multiple, it is read against the company's own history rather than against a nominal threshold.

The four multiples below differ in exactly one thing: the denominator. Price to earnings asks about accounting profit, price to sales about revenue, price to book about assets on the balance sheet, price to operating cash about money that actually reached the bank account. Four questions, four answers. Then one reading says the shares are expensive, another says the opposite, and the reader starts hunting for a mistake in the spreadsheet.

There is no mistake in the spreadsheet. We think multiples that disagree are the market's normal answer rather than a fault, and that the whole job is knowing which denominator is failing right now. Sometimes a low multiple is simply the honest price of an industry that is shrinking, and then there is nothing in the spreadsheet to find.

What is a PE ratio, and what does it measure?

Take a company that earned 480 million and has 240 million shares. Each share carries 2 dollars of profit. At a price of 36 the multiple is 18; at a price of 54 it is 27. The business is identical in both cases. Only the bill the market writes for the same stream of money has changed.

Turn the ratio upside down and it gets more interesting. At a multiple of 18 the annual profit is 5.6 percent of the price; at a multiple of 10 it is 10 percent. The multiple is an earnings yield written backwards, and only in that form can you hold it next to the coupon on a bond.

Two things this number keeps quiet about, and both are large. It does not say whether the profit will grow, and it does not say whether the profit ever turned into cash. No multiple fills the first gap, because that is a question about the future rather than about a ratio. The second gap is filled by the cash denominator, which arrives halfway down this page.

Before you calculate anything, answer the question that settles more than the choice of multiple: which profit sits in the denominator? Profit from the last closed fiscal year can be more than a year old if the company closes its books in June and you are reading in May. The sum of the last four reported quarters is fresher by construction, and it is my default, because it does not ask me to remember each company's calendar.

Two different values of the same multiple in two places are usually not a contradiction.

They are usually a different denominator.

Why the share price says nothing about the price

Which share is cheaper, the one at 9 dollars or the one at 400? Neither, because the question has no content. That sounds obvious right up until you count how many decisions get made the other way.

A share price depends on how many pieces the same business was cut into. A company worth 6 billion, divided into 150 million shares, costs 40 dollars a piece. The same company divided into 6 billion shares costs a dollar. The owner of either holds exactly the same thing, and the instinct that one of them is within reach costs a beginner more than most arithmetic errors do.

A multiple exists to take that division out of the equation.

So all four ratios carry a per-share quantity in the denominator or, in the whole-company version, market capitalisation on top. The share count cancels, and what remains is the relationship between the price and whatever the business produces.

What remains too is a trap few people mention: the share count does not stand still. A buyback lifts profit per share without a single extra sale, and an issue lowers it without a single piece of bad news. The multiple moves in both cases, while nothing at all changed inside the factory.

Price to sales: the multiple that cannot see margin

Revenue is the hardest line in the income statement to dress up. Not impossible, because channel stuffing and premature recognition exist, but harder than net profit, which sits below a dozen accounting choices. Hence the popularity of the sales multiple, and hence its one disqualifying weakness.

Price to sales cannot see how much of that revenue stays. Take two companies with 800 million of revenue and a market value of 2.4 billion each, so an identical multiple of 3. The first keeps 144 million of profit, the second keeps 32 million. You are paying the same for a dollar of the first company's revenue as for a dollar of the second's, while the first turns it into four and a half times more profit. That is not a subtle difference. Those are two different businesses wearing one label.

So why is it used at all? For two honest reasons. Revenue exists even in a company that does not yet earn anything, so it is the only denominator that survives the building phase of a business. It also swings more gently than profit, because it does not pass through a dozen cost lines with policies of their own. The sales multiple is a calmer chart, and a calmer chart makes it easier to see the levels a company has traded at over the years.

The weakness turns dangerous the moment margin moves. A company whose operating margin slides from 22 percent to 15 gives away nearly a third of the profit in every dollar of sales, and the sales multiple will show it only once the market has repriced the shares. Price to sales is a multiple for businesses with a steady margin and for companies that do not earn yet, and in a firm whose mix of revenue growth has changed it is the worst of the four.

Price to book and the company with no factories

This multiple was once the core of value investing and now describes a shrinking part of the market. The reason is not fashion. It is accounting.

Money spent on a plant lands in assets and drips into costs over a decade or more of depreciation. Money spent on research is expensed as incurred under US rules, while international rules allow part of development spending to be capitalised, and only after strict criteria are met (a comparison of the two standards on exactly this point). Two companies can therefore spend the same billion and report a different asset base and a different profit.

The consequence for the multiple is arithmetic. A firm that spends on engineers has a small denominator, so the ratio comes out high, while nothing about that business is expensive. It runs the other way too: a bank trades near its own book value because its assets are loans and securities carried at roughly what they are worth, so the books and the market are looking at the same thing. Before you calculate this ratio, check what the asset side is actually made of, because half the answer is already there.

There is a second half to the same asymmetry, and it works in reverse. A company that bought somebody else's business records the premium it paid above the value of the assets acquired. The denominator grows, the multiple falls, and not one machine was added. Once that premium passes half of total assets, book value without it is often negative and the ratio stops measuring anything.

One case remains where this multiple still earns its keep: a business that does not exist without its assets. A steel mill, a fleet, a transmission grid, real estate. There book value approximates what the company genuinely holds rather than recording a history of spending.

So the question to ask before using it is a single one: are these assets needed to produce anything?

Price to operating cash: the denominator hardest to draw

Net profit is an accountant's opinion about a period. Operating cash is a bank statement. That is why the fourth multiple, the one with cash from operations underneath, is the most resistant of the four to a creative set of books.

Take a company with 260 million of net profit and 115 million of operating cash in the same year. The difference sits in working capital: receivables rise because customers pay later, and inventory rises because goods move more slowly. The profit multiple looks respectable, the cash multiple looks more than twice as bad, and the second one is telling the truth about that year.

Operating cash has weak spots of its own, and two of them are common. A company that stopped paying suppliers on time reports higher operating cash without earning a penny more. A company that sold its receivables to a factoring house days before quarter end does the same thing in one move. Both are legal, both leave a mark in working capital, and both become visible only when three years sit side by side instead of one.

Then comes the weakness the guides skip: seasonality. A retailer collects most of its cash in the fourth quarter, so a multiple built on data through September shows a picture the company does not carry for nine months of the year. The remedy is dull and effective. Build the denominator from the last four quarters together, never from one, and compare it with the same span a year earlier.

Where each of them breaks

multiple asks about breaks when
price to earnings accounting profit per share the company posts a loss or a profit near zero; the profit carries a one-off item
price to sales revenue per share margin moves; the business mixes segments of very different profitability
price to book book assets per share assets are not needed to produce; an acquisition premium sits in the asset side
price to operating cash operating cash per share cash is seasonal; working capital shifted once and will not repeat

Why calculate four when each of them can be challenged? For precisely that reason.

Each row breaks somewhere else, which makes four multiples a cheap consistency test. When all four say the same thing, you know little beyond what you already knew. When one breaks ranks, you have an address worth visiting, and the reason is usually in the filings rather than in the valuation. The same rule applies one floor up: if a model that discounts a few years of cash flow and then guesses the rest disagrees with the multiple, that too is an address rather than a verdict.

Two holes the guides skip

The first concerns cyclical companies and is the exact reverse of the instinct. In a steel mill, a refinery or a chemicals maker the profit multiple is lowest at the very top of the cycle, because profit is at a record and the market already knows the record cannot be repeated. The reading that looks best therefore sits at the worst possible moment.

A year later profit halves, the price falls by less, and the multiple goes up.

The second hole is simpler still: with a loss, the profit multiple does not exist. A negative denominator produces a negative ratio that compares to nothing, and data screens then show an empty cell or a number that means nothing at all. The same applies to a profit near zero, where the multiple jumps into the hundreds and stops measuring anything. A percentage calculated from a base near zero is worthless, and the rule holds here too.

What to read the number against

The level of a multiple carries no information until you say what you are reading it against. There are three references and each has a different weakness.

The company's own history is the strongest, because you compare one business with itself. It does not exist for a company fresh from its listing, and it fails when the firm has changed character, say by moving from one-off licences to a subscription. A peer group rescues those two cases but demands companies of similar profitability and similar forecasts, which almost never exist. The broad market is the weakest, because it throws business models with completely different revenue repeatability into one bucket.

The usual argument against comparing with an industry is a strong one: if two companies do nearly the same thing, the gap in their multiples is information. Sometimes it is. Before you treat it as information about price, though, you have to rule out that it is information about the business model. A firm with revenue that recurs every month and a firm selling the same thing once every seven years will trade at different levels permanently, not only today, and neither is expensive because of it.

I will not hand you a level below which a multiple counts as low, because I know of no published work that sets one. The calibrations circulating on forums are practitioners' rules of thumb rather than a standard, and every one of them treats some industry unfairly. Anyone promising a single number for every company is selling convenience rather than method.

A number from the market only works with a date, so here it is with a date. On 4 September 2026 the multiple on profits forecast for the next twelve months for the broad US index stood at 19.5. Its five-year average was 19.8 and its ten-year average 19.0 (the review of index earnings and valuation from that day, page 12). On one and the same day the market sat below its own five-year average and above its ten-year one. One number, two windows, two opposite conclusions, and that is the whole lesson about thresholds in valuation.

How to set your own multiple band in a spreadsheet

This takes about an hour and you do it yourself, on one company. You need five years of closing prices and the history of whichever denominator you picked.

  1. Pull daily or weekly prices for five years into the first column.
  2. Put the last four reported quarters of the denominator in the second column, rolled forward quarter by quarter.
  3. Divide the first column by the second in the third. That series is the multiple, rather than a single reading.
  4. Calculate a three-year rolling median and draw it on the same chart.
  5. Mark the level the series repeatedly reached from above and from below. Those are the edges of the band, not the all-time high and low.

You will draw those edges by judgement, and two people will draw them differently for the same company. That does not break the method. It only says the output is a position rather than a number: above the band, inside it, below it. Moving the window from five years to ten moves that position, which is why the window gets written down next to the result, along with the date of the reading.

What to do with four readings that disagree

Nothing dramatic. Write all four down, note what sits in each denominator, and see which one breaks ranks. That one is your question for the filings this evening.

And if none of them breaks ranks? Then you have a company where valuation settles nothing, and the decision goes back where it belongs, to the quality of the business.

Order matters too. A multiple is read last, after the question of whether the business earns and the question of whether it can service its debt. The lowest multiple on the exchange usually belongs to the company nobody wants, and that sentence is worth pinning above the spreadsheet before anyone sorts a universe by one column.

The work belongs in a spreadsheet and there is nothing magical in it. If you would rather start from something already assembled, the valuation section of the Full report puts one company's multiples next to its own history. Behind every number it leaves a link to the document that number came from, and what the report does and does not calculate is described separately.

And if you have time for one multiple only, take the one with cash in the denominator.

Cash is the hardest thing to draw.

Frequently asked questions

It is the share price divided by the net profit that belongs to one share. A company earning 480 million with 240 million shares has 2 dollars of profit per share, so at a price of 36 the multiple is 18. A reading of 18 says you pay eighteen dollars today for every dollar of annual profit, and its reciprocal, 5.6 percent, is that profit expressed as a yield on the price. The number alone says nothing about whether the price is high: that answer comes from where the reading sits against the company's own history.
No level is good or bad on its own, and we know of no published work that draws the line. Two companies in one industry can trade at permanently different levels for years because one sells a subscription and the other sells a transaction. The useful question is where today's reading sits against this company's own median over several years. You set that band yourself in a spreadsheet, and the window you pick changes the answer, which is a property of the method rather than a flaw in it.
The denominator, and what each one ignores. Revenue is far harder to dress up than net profit, so the sales multiple tends to be the steadier line. In exchange it cannot see margin: two companies with 800 million of revenue, one keeping 144 million and the other 32 million, carry the same sales multiple and run very different businesses. Price to sales breaks precisely when margin moves, which is the moment an investor most wants to understand.
Because accounting treats a factory and a codebase differently. Research spending is expensed as incurred under US rules, while international rules allow part of development cost to be capitalised once strict criteria are met. A company that spends on engineers therefore reports a smaller asset base than a company that spends the same money on machines. A high book multiple in that kind of business describes the accounting rules rather than the price.
Because they ask four different questions. Accounting profit, revenue, book assets and operating cash separate exactly when something is happening inside the business: margin moves, profit stops converting into cash, an acquisition adds a premium to the asset side. The disagreement is a pointer to where to read next, not an arithmetic error. Four readings that agree usually carry less information than one reading that breaks ranks.
It means nothing without a reference and a date. A low multiple is often the honest price of a shrinking market, of a one-off gain on an asset sale, or of the top of a cycle, where the multiple looks best just before profits fall. A statement with a reference reads differently: below this company's own five-year median, as of a named day. That sentence can be checked; a bare adjective about price cannot.
  • fundamentals
  • research
  • financial-statements
Share:XLinkedIn

Posts are produced with AI tools and go through editorial review by the Taufolio team before publishing. How we write and correct posts

Cookies and analytics

Essential cookies keep you signed in. Analytics runs only if you agree.