How our published lists have done since publication

What the score’s history shows (December 2016 to August 2026)

Hypothetical test: not the results of any account, list or person.

A live record takes years to mean anything, so we also ran a hypothetical test: we rebuilt the past without hindsight about prices or filings. For every month from December 2016 to February 2026 we recalculated a version of the Equity Rank Score for about 2,600 US stocks, using only company filings and prices available on that date, and kept companies that were later delisted, acquired or went bankrupt. Each month we took the 10% of stocks with the highest score (about 420) and followed them for the next six months, after estimated trading costs. The last six-month period ended in August 2026. The comparison is with the average of all scored stocks, not with the S&P 500.

What it showed

• On average, the top 10% did about the same as the average scored stock. The difference was −0.02% a month, too small to tell apart from zero. In this hypothetical test, from December 2016 to August 2026 the scored stocks as a group grew about 14% a year (before costs) and the top 10% about 13.6% a year (after estimated costs), with similar swings: the largest peak-to-low fall was 36% for the top 10% and 32% for all scored stocks.
• The typical stock in the top 10% did somewhat better than the typical scored stock: a median six-month return of 4.9%, against 4.2% for all scored stocks (higher in 56% of months). But the lowest-scored 10% held a small number of very large gains: their average six-month return was 10.2% even though their median was only 1.8%. An equal-weighted group earns the average, not the median, so the two effects cancelled out. The figures in this point are before trading costs.
• No single part of the score showed an advantage that survives the statistical corrections needed when many ideas are tested. The price-trend component came closest.

Limits of this test

• It is a hypothetical rebuild, not the result of any account or person. Real results would differ, for example because of taxes, account fees and the prices actually obtained.
• It uses a simplified version of the score built from historical filings; the live score uses more inputs and somewhat different weights.
• The comparison is with the equal-weighted average of all scored stocks, which gives small companies far more weight than the S&P 500 does. It is not a comparison with the S&P 500.
• It covers one period, mostly rising markets. At least one weighting choice in the tested score was made after looking at this same data, which, if anything, makes the result look better than a truly fresh test would.

That is why every published list is now recorded on its date and measured going forward on this page, and why we show the results whichever way they go. Past results do not indicate or predict future results. Nothing here is investment advice.