What is a good P/E ratio? The answer is another number

Type "what is a good P/E ratio" into a search engine and every answer says the same thing: between 15 and 20, lower is cheaper, compare within the industry. That was written for a market that no longer exists. Here is the one that does.

We cover 968 companies. On 21 September 2026, 180 of them lost money last year, so they have no P/E ratio at all. Of the 788 that earned something, the middle one trades at 25 times its earnings. A quarter trade below 17, a quarter above 39, and 130 of them above 50. A P/E of 15 does not make a company "fairly priced" today. It puts it in the cheapest fifth of the market, and the market usually has a reason.

So the honest answer to "what is a good P/E" is: compared to what? There are three comparisons that turn the number into a judgement, and four ways it lies. That is this post.

What the number is

Price per share divided by earnings per share. If Walmart trades at $107 and earned $2.73 a share last year, its P/E is 39: you are paying 39 years of last year's profit for the right to own a slice of it. Target, in the same aisle, trades at 19.5 times. Same customers, same decade, half the price per dollar of profit. The number is not telling you Target is better value. It is telling you the market expects Walmart's profits to grow and Target's not to, and asking you whether you agree.

That is the whole use of a P/E: it is the market's growth forecast, written as a single number. Low means "not much expected". High means "a lot expected". "Good" depends on whether the expectation is right, which the number cannot tell you.

Comparison one: the company against its own past

The cheapest way to read a P/E is against the same company ten years ago.

Apple traded between 12 and 19 times earnings every year from 2015 to 2019. Since 2020 it has been between 24 and 38, and sits at 45 today. The company did not change its business; the market changed its mind about what a phone maker with a services arm is worth. Anyone holding the "15 to 20" rule sold Apple in 2020 and has been waiting for it to come back for six years.

Walmart: 14 to 17 times from 2015 to 2017, then 22 to 44 every year since 2018, 39 now. Same store, twice the price per dollar of profit. (Two of those years, 2018 and 2019, were charges, not enthusiasm; the ones since are enthusiasm.)

Home Depot: 19 to 28 every year for a decade, 21 today. That is what a company looks like when the market's opinion of it has not moved. If you read the balance sheet post, that stability is the multiple's version of the same story.

Exxon: 8 to 21 in most years, no P/E at all in 2020 because it lost money, 48 in 2016 because it barely earned anything. Cyclical companies have P/Es that swing with the oil price, and a low one often arrives at the top of the cycle, when earnings are fat and about to shrink. That is the point of why cheap is a forecast.

So the first question is never "is 25 high", it is "is 25 high for this company". Our company page draws the price against what the same earnings would be worth at the company's own usual multiple, for exactly this reason.

Comparison two: the company against its industry

Different businesses carry different multiples for good reasons, and the spread is wider than the guides admit. From our tables, the middle P/E of each sector on 21 September:

Source: our price and financials tables, last close to 21 September 2026 against each company's last full-year diluted earnings per share; companies with a loss excluded (they have no ratio).

Read the top and bottom rows. JPMorgan at 17.5 is on the dear side for a bank (Wells Fargo is 13.8, Bank of America 15.2). Microsoft at 27.5 is on the cheap side for technology, where a quarter of companies trade above 76. The same number, 20, would be expensive in one row and a bargain in another. A P/E without its sector is a temperature without a unit.

Two sectors deserve a note. Utilities cluster tightly (18.5 to 23.9 for the middle half) because their profits are regulated and their growth is slow and known; an outlier there is worth a look. Technology spreads widest because it holds both the cash machines (Microsoft, Apple) and the companies priced on what they might earn (Palantir at 282 times, AMD at 211). Tesla, filed under consumer cyclical, is the same animal in a different row, at 337.

Comparison three: the number against the growth

The last comparison is the one that decides. A P/E of 45 is a bargain if earnings double next year and a disaster if they halve, and the ratio looks identical in both cases the day you buy.

Micron trades at 134 times its last full year's earnings, $7.59 a share for the year to August 2025. Its most recent quarter alone, to May 2026, earned $24.67 a share, more than three times that whole year. On the last four quarters the ratio is 23. So the "E" in the 134 is stale, and the market has priced the earnings it expects to be filed next, not the ones that were. That is what a very high P/E on a growing company usually is: an old denominator. It is also what a very high P/E on a company that has stopped growing is, one year later, and from the outside the two look the same on the day you buy.

Which is why I wrote the Comfort Systems post: a 27x stock made of a 7x business and a 4x multiple. The multiple, 14 to 54, is the crowd's forecast. The earnings are the fact. A good P/E is one where the fact has a decent chance of catching up with the forecast, and you only find that out by reading what the company said about next year, not by staring at the ratio.

The four ways it lies

1. A loss has no P/E. 180 of our 968 companies lost money last year, including Ford, Rivian and Snowflake. They do not have a high P/E or a low one; the ratio is undefined, and screeners either show a negative number or a blank. A negative P/E is not "cheap". Sort by P/E and every loss-maker disappears from the list, which is one reason the list looks safer than the market.

2. A one-off year breaks it. Coca-Cola's P/E in 2017 was 158, not because the shares went wild but because a tax charge cut that year's earnings to almost nothing. AbbVie is at 112 today for the same reason: acquisition-related charges, not a collapse in the business. Before you call a P/E high or low, look at whether the "E" was a normal year.

3. "Last year" and "next year" are different ratios. Every P/E on a website is either trailing (last twelve months of reported earnings) or forward (an analyst's estimate for the next twelve), and they can differ by half for a fast-growing company. The ratios in this post use the last full filed year; the app also shows the quarter-by-quarter one. Forward P/Es are cheaper and less true.

4. A low P/E is a forecast too. Verizon at 11.8 is not a mistake. It is the price of a business the market expects to grow slowly and carry a lot of debt. (AT&T shows 8.4 and is lie number 2 in disguise: its 2025 earnings carry the gain on selling DIRECTV; on a normal year it is about 11.) Sometimes the market is wrong, and a low P/E is the whole return. More often it is right, and the P/E stays low for a decade while the dividend does the work. Low is not a bargain. It is a prediction, and you have to disagree with it for a reason.

So what is a good P/E ratio

One that is low against the company's own history, ordinary for its sector, and attached to earnings that were a normal year and are about to grow. All four at once is rare, which is why it is worth looking for.

Here is the routine, four steps:

1. Check the earnings were normal. Open the last annual report; if net income moved by more than a third from the year before, find out why before you trust the ratio.
2. Put the P/E against the company's own ten-year range. Below its usual band with the business intact is interesting. Above it needs a story.
3. Put it against the sector's middle from the table above. 20 is cheap in technology and dear in banking.
4. Ask what growth the number assumes. At 45, the market expects a lot. Read what the company said about next year and decide whether you believe it more or less than the crowd does.

Every company page in our app shows steps 2 and 3 on one chart, the price against what its earnings would be worth at the company's own usual multiple and at its industry's, with the income statement for step 1 below it, and the free report on any company shows you the same page. Step 4 is still you and the filing.

Educational and proprietary. This explains what our research does, not the exact formula behind it, and it is not personalised investment advice. See the full disclaimer.