quantlab.portfolio.predefined.topk_dropout#

Qlib’s TopkDropout: hold topk names and replace at most n_drop of them per rebalance.

TopkDropoutConstructor follows qlib’s TopkDropoutStrategy (qlib/contrib/strategy/signal_strategy.py, method_sell="bottom", method_buy="top", only_tradable=True): turnover per rebalance is capped at about n_drop / topk of the book, and kept names are not rebalanced, where TopNConstructor rebuilds an equal-weight book on every rebalance.

Classes