Gross profitability
What it measures¶
Gross profit is revenue minus the direct cost of producing what was sold (the cost of revenue: materials, factory labour, the servers a software service runs on). It is profit before the costs of selling, research, management and interest. Gross profitability divides it by total assets: how much gross profit each unit of what the company owns produces in a year.
The check compares the latest year's figure with two lines: the sector median — the middle value of similar companies in the sector benchmark — and the company's own five-year median, the middle value of its own previous five years.
Why it is on the page¶
Robert Novy-Marx found that gross profitability tells strong businesses from weak ones about as well as how the shares are priced does. Because it sits high up the income statement, accounting choices further down have little effect on it. It is a level check: what is normal differs greatly between a supermarket and a software company, so the company is compared with its sector first.
Thresholds, their source and the population they were fitted on¶
| Result | When |
|---|---|
| ✓ passed | gross profit ÷ total assets is at least the sector median |
| ✕ failed | it is below the sector median |
| ○ not evaluable | no sector benchmark is stored for the company's group, or gross profit is not stored |
The own-history reading — at least the company's own five-year median — is shown beside the verdict, and the evidence panel keeps the states of all three bases.
- Source: Novy-Marx (2013), "The other side of value"; sector medians from Aswath Damodaran's January industry tables.
- Fitted on: US companies 1963–2010, financial companies excluded.
- Evidence grade: peer-reviewed and replicated (the line); practitioner convention (the sector medians).
- Basis: sector band, with own history beside it.
A worked example¶
Honeywell International, fiscal year 2025: gross profit was 18.8% of total assets, against the company's own five-year median of 19.1% — just under it. No sector benchmark was stored on 25 October 2026, so the check reads not evaluable — no sector benchmark and is left out of the count.
Apple, the same year: 54.3%, above its own five-year median of 49.0%; it too waits for a sector benchmark.
An illustration, not a stored figure: had the sector median for Honeywell's group been 15%, its 18.8% would have passed the sector line while sitting below its own history.
What “No verdict” means here¶
The check is not evaluable in two cases, each named: no sector benchmark is stored for the company's group (the benchmark climbs from the industry to the sector to wider groups until it finds one with at least ten companies), or the company reports no cost of revenue, so no gross profit can be formed. It then leaves the area's count.
What the verdict is and is not¶
A pass says the company's gross profit per unit of assets was at least the middle value of its sector in the latest year. It is not a ranking among those companies, not a forecast, and not an instruction to trade the shares. A company that owns few assets — a software business — can read very high on this figure for that reason alone.