What it is
Return on invested capital (ROIC) measures how efficiently a company turns the money invested in it — debt and equity — into operating profit. In plain terms: for every dollar of capital the business puts to work, how many cents of profit does it earn back each year?
ROIC = after-tax operating profit ÷ invested capital (debt + equity)
If a company earns 20 cents of operating profit for every dollar of capital, its ROIC is 20%. The comparison that matters is ROIC versus the company's cost of capital — roughly what it pays to fund itself. A business earning above its cost of capital is creating value with every dollar it reinvests; one earning below it is quietly destroying value no matter how fast it grows.
ROIC = profit earned per dollar of capital employed
Why it matters
ROIC is the closest thing to a single number for business quality, for one powerful reason: over the long run, a stock's return tends to converge toward the ROIC of the underlying business. A company that reinvests its profits at a 20% return compounds intrinsic value far faster — and more reliably — than one reinvesting at 6%.
It also cuts through the noise that fools simpler metrics. High growth funded at low returns on capital isn't creating wealth; it's a treadmill. ROIC tells you whether growth is actually worth having. The best businesses combine a high ROIC with the ability to reinvest lots of capital at that same high rate — the engine behind every great compounder.
How we actually score it
Most explanations stop at "high ROIC is good." Here is precisely what the model does with it, because a threshold you can't see isn't a standard.
ROIC carries a weight of 1.7 in the quality pillar — the heaviest single input we use. Nothing else in quality counts for more. The score is banded, not linear:
| ROIC | Score contribution |
|---|---|
| 0% | 0.0 |
| 6% | 0.4 |
| 10% | 0.6 |
| 15% | 0.8 |
| 20%+ | 1.0 |
Two things follow from that shape. A business at 6% is not scored as "slightly worse" than one at 20% — it earns less than half the credit, because 6% is at or below most companies' cost of capital and is therefore closer to value-neutral than to good. And the scale saturates at 20%: a company earning 45% on capital scores the same as one earning 20%. Very high reported ROIC is more often a sign of a small or distorted denominator than of a business four times better than an excellent one, so we decline to reward it further.
Durability counts separately
A single year of high ROIC is cheap. Cyclical peaks produce it, one-off gains produce it, and a favorable comparison period produces it.
So the model scores a second, separate leg: the worst annual ROIC over roughly five years, at a weight of 0.6. It asks not "how good was the best year" but "how bad was the bad one."
| Worst year ROIC | Score contribution |
|---|---|
| −5% | 0.1 |
| 0% | 0.3 |
| 8% | 0.6 |
| 15% | 0.85 |
| 20%+ | 1.0 |
This is the difference between a compounder and a flash. A business that never dipped below 15% through five years — including whatever the economy did in them — is telling you something structural about its position. One that averaged 18% but touched −4% in a bad year is telling you something else, and the average alone would hide it.
Why it's harder to measure than it looks
The formula is simple. Getting a trustworthy number for 1,800 companies is not, and we'd rather show the seams than pretend otherwise.
We compute ROIC from SEC filings directly rather than accepting a vendor's figure — after-tax operating profit over debt plus equity, taken from the filed statements. In August 2026 we audited how often that actually resolved, and the answer was uncomfortable: ROIC was missing for 294 of 1,824 scored companies. For the heaviest input in the quality pillar, that is a lot of silence.
Most of it was not a limit of the data. It was our own code.
The invested-capital calculation required book equity to be positive before it would add debt to it. But invested capital is debt plus equity, and that total is routinely positive even when equity is negative. The check was both redundant — the next line already required the total to be positive — and wrong.
The companies it silently dropped are the ones the mistake makes most instructive:
Every one of those carries negative book equity, and in most cases because of years of buybacks — the company has returned more to shareholders than its accumulated book value. That is a capital-allocation choice, not distress. These are exactly the durable, cash-generative businesses a quality screen exists to find, and a one-line guard was excluding them from the metric that would have ranked them highest.
The guard is fixed. We're keeping the account of it here because it illustrates something real about ROIC that a definition can't: the denominator is where the difficulty lives. Invested capital sounds like a fact and is closer to an interpretation — book equity distorted by buybacks, operating profit that needs deriving when a filer doesn't tag it, and a tax rate that has to be assumed. We apply a flat 21% to operating profit to get the after-tax figure, which is a simplification and worth knowing about when you read the number.
What "good" looks like
- ROIC comfortably above the cost of capital — a spread of several points or more. On our scale, 15% earns most of the available credit and 20% earns all of it.
- Consistency over peak — a high ROIC sustained through a bad year beats a higher one that wasn't. That's why the worst-year figure is scored separately rather than folded into an average.
- A reason it persists — durable high ROIC almost always rests on a moat: brand, network effects, scale, switching costs. Without one, competition drags returns back toward the cost of capital.
- Room to reinvest — a high ROIC is worth most when the company can deploy more capital at similar rates, not merely harvest a small saturated niche.
Where to see it
Stock pages show ROIC with a letter grade in the quality signals panel, sourced from filings and refreshed each weekday with the rest of the fundamentals. Where it can't be computed, it's left off rather than replaced with an estimate — after the audit above, we would rather show a gap than a number we can't stand behind.