Our stock score was proven on the S&P 500. The obvious question: does it work below the mega caps?
We rebuilt the whole test — as-filed SEC fundamentals, signals dated by filing date, monthly top-20
portfolios — on the S&P 400 and 600.
Result
Top-20 excess over the same pool’s equal-weight baseline: +7.9%/yr (t 2.4), and in the
well-covered 2019+ era +13.6%/yr (t 2.4). Against 1,000 random 20-name portfolios drawn from the
same pool, the real strategy beat all 1,000. Clipping each stock-month at +100% barely moved it —
it is not a lottery-ticket artifact. Drill-down: the edge concentrates in the smallest index (S&P 600),
exactly where markets are least efficient.
Where the edge stops
We pushed one level further: the ~4,500 listed names outside the S&P 1500 entirely. The raw signal still
ranks that pool — but the winning portfolio lives in sub-$150M microcaps the fund cannot actually trade.
Apply the fund’s real liquidity and price gates and the edge drops to +5.0%/yr at the 92nd–95th percentile
of its null — below our adoption bar (beat the 95th, ideally the max). The edge dies at the S&P 1500
boundary.
Caveats, honestly
No membership history exists for these indexes, so the universe is survivorship-biased — we measured
the inflation at roughly 5%/yr on absolute returns, which is why only the excess-vs-own-pool and the null
test are treated as evidence. Adopted with constraints: a cap on how many small-cap slots the book may hold,
and sizing that respects the worse drawdown profile (backtested at -44% to -50% in that sleeve).