What this is. This page reports a measured
state, not a recommendation. A name appears
when its RS line (close ÷ index) and its closing price are both at a 252-session high on the same day.
What the evidence says. Measured in R under a 10%
stop and a 21-week-MA trail armed from entry,
against a
control that holds every gate except the
RS-line condition, clustered by entry month:
US in-sample edge
+0.108 (t 1.93), out-of-sample
+0.067 (t 1.71).
What it does not say. This is a
marginal result — both t-statistics sit just above the
1.5 threshold. It replicates in Thailand and the US and
fails in Japan and Hong Kong. An earlier
published figure for the same rule was about 60% too large because it measured dividend-adjusted prices
against a dividend-free index; the numbers above correct that. The underlying research file classes it
as
candidate, not validated. It is measured on a dataset in which no symbol delists before 2024,
so the relative edge is reported and the absolute return is not.