Earnings per share growth fell for 312

On 2026-09-07, of the 1,459 US listings holding both a five-year revenue compound growth rate and a five-year earnings per share compound growth rate, 312 compounded revenue upward and earnings per share downward. Revenue growth belongs to the company; earnings per share growth belongs to the owner, and the two can point in opposite directions for five years at a time.

Cross the signs:

Counted with POST /api/screener/search on 2026-09-07. The public endpoint reports at most fifty matching rows, so a query that comes back under fifty is reporting a true count. The pool was split by sector and then by market capitalisation until every part came back under fifty, and the parts were added. The four cells sum to 1,454; the remaining five listings sit at exactly zero on one of the two rates.

312 of 1,459 is 21.4 percent. Restricted to the 1,324 that grew revenue at all, it is 23.6 percent: close to one in four.

Half of it is the share count

Earnings per share is earnings divided by shares. There are two ways for it to fall while revenue rises: earn less on more revenue, or divide by more shares.

The split is almost exactly even. Dilution accounts for about half of the divergence and cannot account for the other half, where the share count went down and earnings per share still fell. In that half the company bought its own stock back, which pushes earnings per share up, and the figure fell anyway. So something in the income statement between revenue and net income moved further than the buyback could offset: margin, interest, tax, or a mix.

Those are two genuinely different situations wearing one headline. A company issuing equity to fund growth that has not arrived yet is early. A company buying back stock while its earnings fall faster than the share count is spending cash to slow the appearance of a decline. Neither is visible in a revenue growth rate, and both are visible the moment the second rate is put next to the first.

The denominator is the most important number on this page

1,459 listings out of a pool of 5,257 on 2026-09-07. This study covers 28 percent of the market as it stood that day, and the missing 72 percent is not missing at random.

A five-year revenue growth rate is held for 3,532 listings. A five-year earnings per share growth rate is held for 1,492. The gap between those two numbers is the whole story of what a compound growth rate can and cannot say about earnings.

Two rules do the work. The figure is withheld when fewer than five filed years exist, which removes every recent listing. And a compound growth rate is meaningless when either endpoint is negative: a company that lost 2 dollars a share five years ago and earns 1 dollar today has no compound rate, because there is no positive number to compound from. The arithmetic would produce a figure, and the figure would be nonsense.

So the population here is companies that were profitable five years ago and are profitable now. That is a considerably better-behaved set than the market, and it means 312 is a floor. The companies whose earnings crossed zero in either direction are not in the count, and by any ordinary reading they are the more dramatic cases.

What an endpoint-to-endpoint rate hides

Both rates are computed from two points, five years apart. Everything between them is invisible.

A company that grew revenue steadily at 8 percent for five years and a company that grew 40 percent in year one, collapsed in year three and recovered report the same five-year rate. The registry says so in its own caveat on the figure, and it is the reason the same measure is published at three years as well: two windows on one series, where the gap between them dates the inflection.

The share count comparison above has a sharper version of the same problem. The share count change is a one-year figure and the growth rates are five-year. A company that diluted heavily three years ago and has been buying back since appears in the "share count down" column, which is true about last year and misleading about the period the earnings per share rate covers. The even split above should be read as an indication that both mechanisms are common, not as a decomposition of the five years.

Where to look next on a company that is in the 312

Three checks, in order, and all of them are in the filings.

The share count line at the bottom of the income statement. Diluted weighted average shares outstanding, five years of it. That settles the dilution question directly rather than through a one-year proxy.

Operating margin against revenue. If revenue rose and operating margin fell far enough, the operating line is flat or down before any financing enters, and the answer is in the cost base rather than the capital structure.

Interest expense. Revenue growth funded by borrowing shows up as a rising interest bill, which lands below operating profit and above net income, which is exactly where a divergence like this hides.

What this leaves out

Counts are of listings in the pool as it stood on 2026-09-07, not deduplicated companies, and they include listings whose figures the product would refuse to display on their own company pages.

Both rates are endpoint-to-endpoint compound annual growth over five filed fiscal years. They say nothing about the path between the endpoints, and a company whose fiscal years do not align with another's is being compared over a different five years.

The earnings per share rate is withheld where either endpoint is negative, because a compound rate over a sign change is not a rate. That exclusion removes the most severe cases from this study in both directions.

The share count figure is a one-year change and the growth rates are five-year. The comparison between them is indicative and is not a decomposition.

No company is named and nothing here is a view on any security. A falling earnings per share is not evidence of anything on its own, and the three checks above are where the evidence would be.