ROIC vs. ROE: the metric a quality screen actually needs
Return on equity gets inflated by debt; return on invested capital doesn't. Here's why a quality screen leans on ROIC, with the exact thresholds involved.
Return on equity is the profitability ratio most investors learn first, and it has a structural flaw most of them never unlearn: it can be improved without the underlying business getting any better. Load a company up with debt, buy back shares with the proceeds, and equity shrinks while net income barely moves — ROE climbs, and nothing about how well the company turns capital into profit has changed. Return on invested capital (ROIC) exists to close that loophole, and it's worth walking through exactly how, and where it still breaks.
The mechanical difference
ROE divides net income by shareholders' equity. Equity is whatever is left after debt, so ROE is a levered number — the same operating business can show a wildly different ROE depending on how it's financed. ROIC instead divides after-tax operating profit (NOPAT) by invested capital: debt plus equity minus excess cash. Debt goes into the denominator instead of being netted out of it, so a company can't flatter its ROIC by simply borrowing more. It has to actually generate more profit per dollar of capital employed, regardless of whether that capital came from bondholders or shareholders.
That's the textbook version. What's more useful is watching what happens when you build a real screen around each one.
Where this shows up in a live screen
Zorro's screener runs eight one-click presets against a universe of over 3,400 US stocks, and the Quality compounder preset is a direct test of this idea. Its actual filter, unedited:
- Market cap ≥ $1B
- ROIC ≥ 15%
- Operating margin ≥ 15%
- Revenue growth ≥ 5%
- Debt/equity < 1
- Net margin > 0
Notice what's doing the real work: the ROIC floor and the debt/equity ceiling are two separate constraints, not one. A screen that used ROE alone could pass a heavily levered company on profitability grounds and separately fail to notice the leverage, because the leverage is already baked into the number that looks good. Requiring both ROIC and a debt ceiling means a company has to clear the capital-efficiency bar on an unlevered basis, then clear a second, independent bar for how much debt it's carrying to get there.
The site's underlying ranking model — the one that generates the screener's ranked value picks — makes the same substitution more explicitly. Its documentation notes that ROE and ROIC "rank together at Spearman 0.81 but not identically," and keeps both: ROIC carries more than double the weight of ROE in the composite, precisely because it's the cleaner of the two capital-quality signals. When that model's factor set was rebuilt in August 2026 to correct for a sector bias — the old weighting had let 38 of the top 50 ranked stocks be financials, roughly three times their eventual base rate of 12 of 50 — adding ROIC into the mix (at 98.8% data coverage across the universe) was one of the two changes that fixed it. The top 50's median ROIC went from 8.92%, barely above the whole pool's 8.43% median and therefore no quality edge at all, to 15.55% under the corrected weights.
Why the raw number still needs a leash
ROIC solves the leverage problem, but it introduces its own: the denominator can go tiny or negative for reasons that have nothing to do with capital efficiency — a company that's bought back almost all its equity, one coming off a write-down year, or a data error in a single quarter's balance sheet. Divide a normal profit figure by a near-zero or negative invested-capital base and the ratio can print numbers like -3,215% or several thousand percent, which is a data artifact, not a signal. Zorro's screener clamps every ROIC value to a -50% to +60% range before it's used in ranking or display, for exactly this reason: percentile ranking is monotonic, so one absurd outlier can't inflate the stock that owns it, but an unclamped one would still shove every other stock's rank down a slot to make room for a number nobody should trust. The same logic is why the "High ROIC" badge on individual picks only fires at 20%+, roughly the value pool's 85th-90th percentile — high enough to mark a stock out rather than describe half the list.
Reading ROIC against the rest of the picture
None of this means ROIC alone tells you a company is a good investment — a high ROIC business trading at 40 times earnings is a different proposition from the same business at 15 times, which is why the screener's Quality compounder preset pairs the ROIC floor with growth and margin filters rather than running on ROIC in isolation, and why a companion piece on free cash flow yield covers the cash-conversion side of the same question — a business can post a strong ROIC on paper while its reported profit and its actual free cash flow diverge for reasons worth checking separately. Recently listed companies are a useful stress test for all of this at once, since many haven't posted four full quarters of comparable financials yet; the recent IPOs page shows which names in that cohort already have enough trading and reporting history for these ratios to mean something, and which are still too new to screen on invested-capital returns at all.
The broader point holds past ROIC specifically: any ratio built by dividing one number by another is only as trustworthy as its denominator, and the more a company's financing choices can move that denominator around, the more a screen has to correct for it rather than take the printed number at face value.