No negative P/E ratios, 2,218 losses

A count of listings carrying a negative P/E ratio in this pool on 2026-09-07 returned zero, and on the same day 2,218 of the 4,961 listings holding a trailing earnings yield had a negative one. The zero is not a property of the US market. It is a rule: the ratio is withheld whenever trailing earnings are not positive.

Two sites can show the same company with a P/E of minus 8 and a P/E of nothing at all, and the second is right. A price to earnings ratio with a negative denominator is not a low ratio. It is an undefined one, and printing it as a number puts a company that lost money into the same column as a company that made money cheaply.

The interesting question is how much of the market that removes.

Nearly half the listings we can measure lost money

The earnings yield is the figure that survives a loss, because a negative numerator produces a negative yield rather than an undefined one. On 2026-09-07 the screener held one for 4,961 of the 5,257 listings in its pool.

Two thousand two hundred and eighteen listings reported negative trailing net income. That is 44.7 percent of the listings for which the figure is held and 42.2 percent of the whole pool.

The figure was obtained by partitioning the non-negative range into intervals until every one returned an exact count, summing to 2,743, and subtracting from the coverage figure. It is a measurement rather than an inference.

Which is why the P/E column is blank for 2,897 listings

The P/E is held for 2,360 of the 5,257, or 44.9 percent. The other 2,897 have no ratio.

The 2,897 account for themselves exactly:

Those four numbers were measured separately and they sum to the coverage gap without a residual, which is the check that the explanation is complete rather than plausible.

Read the two coverage figures side by side and the shape of the US market falls out of them. The earnings yield is held for 94.4 percent of the pool. The P/E is held for 44.9 percent. The gap is not a data gap. It is the proportion of listed US companies that did not report a profit in their last trailing year.

The 342 that made money and still have no P/E

Subtract the two and something is left over. There are 2,702 listings with a positive earnings yield and 2,360 with a P/E, a difference of 342.

Those 342 reported positive trailing net income and carry no price to earnings ratio. The reason is that the two figures are not the same measurement, and this is worth knowing before treating one as the reciprocal of the other.

The earnings yield here is filed net income over market capitalisation: a whole-company figure. The P/E prefers a data vendor's trailing ratio of price over earnings per share, and falls back to our own price over filed earnings per share when the vendor publishes none. Net income and earnings per share attributable to common shareholders are different lines, separated by preferred dividends and by the share of profit belonging to minority interests. A company can report positive net income and no positive earnings per share.

The P/E is additionally refused whenever our own filed earnings per share is zero or negative, even where a vendor ratio exists and is positive. That rule is why the count of negative P/E ratios is zero rather than small.

Two clamps worth knowing about

The earnings yield is clamped to a range of minus 200 to plus 200 percent, so that a share priced near zero cannot produce an absurd figure. The clamp is doing real work.

Fifty-one listings sit exactly at the plus 200 ceiling, which means their filed net income was at least twice their market capitalisation. At least 273 sit exactly at the minus 200 floor, where the trailing loss was at least twice the market capitalisation; the true figure is higher, because the count in one sector ran past the endpoint's reporting ceiling.

Both piles are real companies with real filed figures, and both are stacked at a boundary the product chose rather than at a value the market produced. A distribution of earnings yields is not a smooth curve at the ends, and any average over it is affected.

What to use when the P/E is blank

The earnings yield, with its sign. It ranks loss-makers below profit-makers instead of removing them, which is the behaviour you want when the question is which companies earned anything at all.

Price to sales, when nothing on the earnings line is usable. It is held for 4,333 listings against the P/E's 2,360. It says nothing whatever about whether the revenue is profitable, and that is not a small caveat, it is the whole caveat.

A blank, on purpose. The most useful thing a screener can do with an undefined ratio is show nothing. A blank cell cannot be sorted into a ranking, cannot be averaged, and cannot make a company that lost money look like a bargain.

What this leaves out

Every count is one reading of the public screener on 2026-09-07, against prices stamped 2026-08-26. The counts move nightly.

The sign of the earnings yield is the sign of filed net income, so the 2,218 is a count of listings whose trailing net income was negative. It is not a count of companies: a company with two share classes appears twice. It is also not a count of the market, because 296 listings hold no earnings yield at all and are absent from both sides of the table.

The count at the minus 200 clamp is a floor. One sector's count ran past the endpoint's ceiling of 50 and is reported as at least 50 rather than guessed.

The explanation offered for the 342 listings with a positive earnings yield and no P/E follows from how the two figures are defined and from the documented refusal rule. It is not a per-company diagnosis, and some of the 342 will have a different cause.

Trailing means the last twelve months as the source measures them. A company that has since filed a profitable quarter is still counted on the figure held on the day of the reading.

Nothing here is a view on any company. A blank P/E is an accounting fact about a period, not an assessment of a business.