A sector-relative score is only as useful as your evidence that it helps you. The cheapest evidence is a log: the score, its five pillar values, the calculation date, the peer group size, what you did, and what happened relative to the sector afterwards. Forty entries is about the point where the log starts to say something.
Sort by total first
If the high-score entries did not outperform the low-score entries relative to their sectors, either the weights do not fit your holding period or the holding period does not fit the weights. That is a finding about you and the tool together, and it is worth more than any single reading.
Then sort by pillar
Most people find that one pillar carried the result and one contributed nothing for their particular style. A long-term holder often finds quality and safety did the work; a shorter-horizon trader often finds momentum did. That is the only honest reason to reweight, and it should be done by writing the new weights down before the next batch of entries, so the next forty are a real test rather than a fit.
Judge outcomes against the sector
A stock that fell five percent while its sector fell fifteen was a relative win, and a score that pointed to it did its job. Judging against the index would score that entry as a loss and teach the wrong lesson. Sector-relative scores need sector-relative outcomes.
Keep the shape, not only the sum
Two entries with a total of 70 can have opposite pillar profiles. If only the total is logged, the log cannot later show that lopsided profiles behaved differently from balanced ones, which is often the most useful thing it has to say.
The pillar definitions and default weights are documented on the Stock Expert AI methodology page. Live sector-relative scores for US-listed stocks are at www.stockexpertai.com.
This is educational content, not investment advice. Past performance does not guarantee future results.
No comments:
Post a Comment