A sector-relative score converts every metric into a percentile inside the company's own sector. That works well when the sector has hundreds of members. It works badly when the "sector" is a dozen companies.
What goes wrong #
With 12 peers, one company moving from 6th to 4th place jumps roughly 17 percentile points on that metric without anything meaningful changing in its business. With 200 peers the same reshuffle moves a company by about one point. Percentiles are only as smooth as the group is large.
Practical rules #
- Look at the peer count before trusting the score. Under 15 peers, read the components, not the total.
- Prefer sub-sector labels that group similar business models; "integrated oil" and "pure refiner" do not belong in one percentile pool.
- Watch deep cyclicals at the bottom of their cycle: they score badly right before earnings recover.
- Check the calculation date; fundamentals arrive with a lag, prices do not.
Where the numbers come from #
The figures are taken verbatim from market data providers and never generated by a language model; the model only writes the commentary. The full method is documented on the Stock Expert AI methodology page.
This is educational content, not investment advice. Past performance does not guarantee future results.