Return on invested capital (ROIC) across this market
1,614 of 5,521 US-listed companies publish a return on invested capital computed after tax from their filed accounts. The middle one earns +6.7% and 412 of them earn less than nothing on the capital they employ.
Every figure below counts companies rather than listings. No company is named, ranked or recommended anywhere on this page.
What the market earns on its capital
- Publishing a return on invested capitalOut of 5,521 in the catalogue.
- 1,614
- Of those, above zero
- 1,202
- Of those, below zeroCapital employed, nothing earned on it. A real figure, not a missing one.
- 412
- Publishing no figure at all
- 3,907
Notes on this count
Across every published figure the quartiles are -0.1%, +6.7% and +13.7%. This is the catalogue’s own denominator, one row per company with a published figure. The return on invested capital benchmarks are computed over a cohort built to a different rule, with extreme values clamped before the percentiles are taken, so the two sets of numbers are close and are not the same thing. Neither is wrong; they answer different questions and each says which.
GET /api/public/companies. Catalogue built Oct 16, 2026. Prices of Oct 15, 2026.
Why 3,907 of them publish nothing
- The filed borrowings do not add up to a total we can stand behind: the filer's own liability subtotals leave room for debt we did not read.
- 1,139
- Not meaningful for this company, such as a P/E on negative earnings or a bank's EV/EBIT.
- 789
- The debt total on file was summed from short-term debt, long-term debt and lease obligations, and at least one of those was never captured. An absent term is spent as a zero inside that sum, and this balance sheet does not show the term is empty, so we will not build a leverage ratio, an enterprise value or a return on capital on it, or rank the company on one.
- 681
- The filed income tax is outside 0 to 100% of pre-tax income (a tax benefit on a profit, a charge on a loss, or a charge above the profit), which is a one-off in the tax line and not a rate operating profit was taxed at, so after-tax operating profit is not computed.
- 529
- No tax rate can be read for this year: the income tax or the pre-tax income line is not filed, or pre-tax income is zero, so after-tax operating profit is not computed.
- 236
- No figures have been computed for this company yet, though we hold its filings.
- 158
- No operating income line is filed for this year.
- 152
- No filed statements on record for this company.
- 115
- No cash line is filed for this year.
- 38
- No shareholders equity line is filed for this year.
- 21
- No income statement for the latest filed year is on file, or the one on file reports neither revenue nor net income.
- 19
- We hold no figure here and no stored reason for its absence. The gap is ours, not the company's.
- 14
- No annual statement of this kind is on file for this fiscal year.
- 9
- A stored zero the filing does not prove, withheld rather than published as if it were a measurement.
- 5
- The balance sheet on file for this year is a copy that names no SEC filing, and its debt total of 0 is that copy's fill, not a figure the company stated. A debt-free balance sheet is not assumed from it.
- 2
Notes on this count
A bank does not have invested capital in the sense the ratio means, and a company with no filed statements on record has no capital we can read. Both are blanks here and both say which kind of blank they are, because a blank rendered as a zero is a company that appears to earn nothing.
GET /api/public/companies. Catalogue built Oct 16, 2026.
Clean books, high ROIC, every condition at once
- Piotroski F-score at or above 7Of the 3,699 publishing that figure.
- 516
- Return on invested capital, after tax at or above 12Of the 1,614 publishing that figure.
- 481
- Price to earnings at or below 30Of the 2,561 publishing that figure.
- 1,723
- Debt to equity at or below 1Of the 2,238 publishing that figure.
- 1,491
- Publishing all 4, so the screen can be run at all4,517 companies cannot be asked.
- 1,004
- Clearing every one of them together
- 53
GET /api/public/companies. Catalogue built Oct 16, 2026. Prices of Oct 15, 2026.
Earnings yield and ROIC, every condition at once
- Enterprise value to EBIT above 0Of the 1,385 publishing that figure.
- 1,385
- Enterprise value to EBIT at or below 12.5Of the 1,385 publishing that figure.
- 274
- Return on invested capital, after tax at or above 15Of the 1,614 publishing that figure.
- 352
- Publishing all 3, so the screen can be run at all4,403 companies cannot be asked.
- 1,118
- Clearing every one of them together
- 103
GET /api/public/companies. Catalogue built Oct 16, 2026. Prices of Oct 15, 2026.
Notes on this count
Both screens are named entries in the discipline catalogue the screener runs, and every level above is read from it rather than typed here, so a level changed there changes this count. The gap worth reading is between the last two rows of each block: most of the market does not fail these screens, it cannot be put to them. A third quality screen is not run at all on this page, because return on equity and five-year revenue growth are not published in the table this page counts. A count over some of a screen’s conditions, printed under the screen’s name, would be a number flattering itself.
The thing this page cannot count
There is no moat score here and there will not be one. A durable advantage is what this reader is buying, and it is not a line in any statement: what the accounts hold is the residue it leaves, a high return on capital that did not get competed away over some number of years. Counting the residue is honest. Scoring the cause would be a number invented to fill a column.
The same applies to the level. Nothing here says twelve percent is the right line, or fifteen: those are the levels two named screens run, and where the line belongs for a particular business is your judgement, not an output.
What this page cannot tell you
A count of the market says nothing about the company you hold, and a return on capital measured once says nothing about the direction it is travelling. The way a quality holding fails is rarely a bad quarter: it is a return on capital that drifts down for three years while the story stays the same, and nobody recomputes it because nothing dramatic happened.
The audit will read companies you already own against these tests at no charge. What is sold is the recomputation, and for a quality reader the guardrail that matters is a return on capital floor: the number that drifts for three years without a bad quarter to mark it.
Every guardrail you have saved is re-tested every night against newly filed accounts, and you get one email the first night one breaks, naming the company, the reading and the filing it came from. What to do about it is yours. It comes with Professional.
The check runs every night at 23:30 UTC. A filing is picked up the night after it is published.
The audit needs no account. A new account starts with 30 days of Professional, no card.
Meeting a test is not a reason to buy. Every count above is read from the public endpoints when the page is built, never typed into it: the whole table is at the company catalogue and every formula is at methodology.
What counts as invested capital is defined there, and it is the whole of this page.