P/E Ratio: The Price of a Dollar of Profit, Worked From Real Filings
The P/E ratio is a stock's share price divided by its earnings per share over the last twelve months. Microsoft closed at $516.17 on 25 Sep 2026 after earning $17.95 per diluted share in the year to 30 Jun 2026, so buyers paid 28.8 times a year's profit.
A share price on its own says nothing about whether a stock is expensive. A $900 share can be a smaller business than a $200 one, and a rising price can mean a company is simply earning more. The P/E ratio fixes the first problem by putting the price next to what each share earned. This page works it out from the companies’ own SEC filings, so every figure can be checked.
How it works
Take the price of one share and divide it by the profit per share. The top of the fraction is the market’s price on a given day. The bottom is earnings per share, a company’s net income for the period divided by its average share count, as printed in its earnings report.
The answer reads as “times earnings.” A P/E of 20 means buyers are paying $20 for every $1 of annual profit behind a share. Turned upside down, the same figure gives the earnings yield: $1 of profit on $20 of price is 5%.
The SEC’s investor site describes it as a gauge, a way to see whether a price is high or low compared with the past or with other companies. It is not a verdict on its own. A high ratio can mean buyers expect earnings to grow fast; a low one can mean they expect them to fall.
Both halves move. The price changes every second the market is open. The earnings change four times a year, when a new quarterly or annual report lands. So most day-to-day movement in a P/E is the price.
Which earnings go on the bottom
Three choices decide the number, and a quoted P/E does not always say which were made.
Trailing or forward. A trailing P/E uses earnings already reported, usually the last twelve months, often shortened to TTM. A forward P/E uses analysts’ estimates for the coming year. Every P/E on this page is trailing, because a forecast cannot be recomputed from a filing.
Diluted or basic. Diluted EPS counts the extra shares that options, stock awards and convertible securities could create; basic EPS does not. For a profitable company diluted EPS is never larger than basic, so it gives the same or a higher, more cautious P/E. This page uses diluted.
GAAP or adjusted. GAAP earnings follow US accounting rules and are what the 10-K and 10-Q report. Many companies also publish an adjusted, non-GAAP figure that leaves out chosen costs. Broadcom’s results release of 12 Jun 2024, for example, gave GAAP diluted EPS of $4.42 for the quarter and non-GAAP diluted EPS of $10.96 in the same line. A P/E on an adjusted figure is often lower, and it is not comparable with one on GAAP earnings.
Building the twelve months. When the latest filing is an annual 10-K, the last twelve months are that fiscal year. When it is a quarterly 10-Q, the TTM figure is last fiscal year’s EPS, plus this year’s EPS to date, minus last year’s EPS for the same months.
A worked example
Microsoft, the simple case. Its fiscal year ends on 30 June. The 10-K it filed on 29 Jul 2026 reports diluted EPS of $17.95 for the year from 1 Jul 2025 to 30 Jun 2026, and no later quarter had been filed by 25 Sep 2026, so those twelve months are the trailing year.
Microsoft closed at $516.17 on Friday 25 Sep 2026. P/E = $516.17 divided by $17.95 = 28.76, which rounds to 28.8. The earnings yield is $17.95 divided by $516.17, or 3.48%.
Apple, the case that needs building. Its last filing was a 10-Q for the nine months to 27 Jun 2026. From its filings:
- fiscal 2025 diluted EPS, the year to 27 Sep 2025: $7.46
- plus nine months to 27 Jun 2026: $6.88
- minus nine months to 28 Jun 2025: $5.62
- equals twelve months to 27 Jun 2026: $8.72
Apple closed at $341.07 on 25 Sep 2026, so its P/E was $341.07 divided by $8.72, or 39.1. Using the fiscal 2025 figure alone would have given $341.07 divided by $7.46, or 45.7: the choice of which twelve months moves the answer by more than six points.
Reading the number against something
A P/E is only useful next to another P/E. Three comparisons are common: the same company in earlier years, other companies in the same industry, and the market as a whole.
The company against itself. This is the cleanest comparison, because the business, its accounting and its share structure stay roughly the same. The Microsoft figures below show how far one company’s ratio can travel.
Against its industry. A bank and a software company earn money in different ways, carry different risks and have different growth, so their ratios sit in different ranges. The eight companies below span banking, retail, drugs, software and chips, and some of the spread between them reflects that.
Against the market. An index P/E is the combined price of all members over their combined earnings. The valuation page covers why price to earnings is only one of several routes to a number, and the one that ignores the balance sheet.
The original data
The data: eight large US companies, one closing price each, and the GAAP diluted EPS in their own SEC filings. Prices are Yahoo Finance closes for Friday 25 Sep 2026. Earnings come from the SEC’s XBRL company facts, downloaded the same day, built into twelve-month figures from each company’s latest 10-Q or 10-K as described above. Each company’s inputs, filing dates and result are in a CSV of the eight P/E ratios.
The results, price divided by twelve-month diluted EPS: Costco $922.77 / $19.88 = 46.4. Apple $341.07 / $8.72 = 39.1. Walmart $107.98 / $2.76 = 39.1. Johnson & Johnson $271.22 / $8.62 = 31.5. Microsoft $516.17 / $17.95 = 28.8. Nvidia $225.07 / $7.91 = 28.5. Amazon $249.67 / $12.43 = 20.1. JPMorgan Chase $343.06 / $23.34 = 14.7.
The highest ratio is more than three times the lowest. Costco and JPMorgan Chase both earn billions a year; the market simply paid 46.4 times a year’s profit for one and 14.7 times for the other. Note the dates, too: Costco’s latest 10-Q runs to 10 May 2026, so its twelve months end four and a half months before the price date, while Nvidia’s run to 26 Jul 2026.
Now one company over ten years. For each Microsoft fiscal year, the chart below divides the last close on or before 30 June by that year’s diluted EPS as shown in the latest 10-K that reports it. For fiscal 2017 that is $3.25, restated when Microsoft adopted a new revenue accounting standard in fiscal 2018; as first reported it was $2.71, which would put that year’s ratio at 25.4. The figures are in a CSV of Microsoft’s P/E by fiscal year.
From fiscal 2017 to fiscal 2026 the ratio was: 21.2, 46.3, 26.5, 35.3, 33.7, 26.6, 35.2, 37.9, 36.5 and 20.8. The fiscal 2026 figure is $373.02 over $17.95, taken on 30 Jun 2026, though Microsoft only published the $17.95 on 29 Jul 2026. By 25 Sep 2026 the same $17.95 sat under a $516.17 price, and the ratio was 28.8.
The inputs, close then diluted EPS: FY2017, 30 Jun: $68.93 and $3.25. FY2018, 29 Jun: $98.61 and $2.13. FY2019, 28 Jun: $133.96 and $5.06. FY2020, 30 Jun: $203.51 and $5.76. FY2021, 30 Jun: $270.90 and $8.05. FY2022, 30 Jun: $256.83 and $9.65. FY2023, 30 Jun: $340.54 and $9.68. FY2024, 28 Jun: $446.95 and $11.80. FY2025, 30 Jun: $497.41 and $13.64. FY2026, 30 Jun: $373.02 and $17.95.
When it fails
One unusual year swamps it. Microsoft’s fiscal 2018 P/E of 46.3 was more than double the year before. The price rose 43% over the year, from $68.93 to $98.61, but the larger part of the jump came from the bottom of the fraction: income tax expense was $19.9 billion, against a restated $4.4 billion the year before, and diluted EPS fell to $2.13 from a restated $3.25. A screen for “P/E under 25” would have dropped the stock that year because of one year’s tax bill.
No earnings, no ratio. When a company loses money, EPS is negative and the P/E is meaningless. Many data sites show it as blank or “N/A”, which quietly removes loss-making companies from a P/E screen.
The earnings are old. A trailing P/E uses the last reported months. Costco’s figure above ends on 10 May 2026, and it went stale the day before the price date: on 24 Sep 2026 Costco filed its full-year results with the SEC, $20.76 per diluted share for the 52 weeks to 30 Aug 2026. That figure reaches the XBRL data only with the 10-K. On it, Costco’s P/E was 44.4, not 46.4. If a business has changed since its last filing, the ratio describes a company that no longer exists.
Mixed definitions. Placing a P/E built on adjusted EPS next to one built on GAAP diluted EPS, or a forward ratio next to a trailing one, compares two different measurements. Check which one a quote page shows before comparing.
Different businesses. JPMorgan Chase at 14.7 is not “cheaper” than Costco at 46.4 in any useful sense. A bank’s earnings swing with credit losses and interest rates; Costco’s rest on store sales and membership fees. Comparing their ratios mostly measures that difference.
Price leads, earnings lag. On the same $17.95 underneath, Microsoft’s ratio was 20.8 at the 30 Jun 2026 close and 28.8 on 25 Sep 2026; the price did all of that. When a stock rallies ahead of a report, the ratio climbs; when the report lands, it can drop overnight. A P/E read in the week before earnings is often about to change.
Related
The valuation page sets price to earnings beside cash flow and asset-based methods, and shows why they disagree. The earnings report page explains the filing where the bottom half of this ratio comes from, and value investing covers the approach that leans on low ratios, along with the traps in it. For the step-by-step method, the how to value a stock guide puts P/E in order with the other checks. The market cap page does the same arithmetic at the level of the whole company.
Before I compare two P/E ratios, I check that both use the same earnings: GAAP diluted, the last four reported quarters. A P/E built on a company’s own adjusted figure, or on next year’s forecast, is a different number wearing the same label.
— Michael Whitman
This page is educational, not financial advice. Test every idea on your own charts before risking money.