quantlab.portfolio.predefined.approximate.top_n#
The approximate backtest’s top-n: every decision bar’s equal-weight book at once.
The implementation quantlab.portfolio.approximate.implementation chooses
for a TopNConstructor; it reads the rule’s config. Each bar’s book is the one
TopNConstructor.construct gives the bar with nothing locked: the same
picks (a stable sort, ties resolved by symbol order), top_n reduced where
fewer symbols can be picked (one warning names those bars), and 1 / k
(0.5 / k per side long-short) on each pick. The tie events are not
counted. The books do not read the holdings.
Functions
Return every bar's top-n book; top-n reads neither the holdings nor a stopping rule. |