A0678
Title: Signal-based subset selection for long-term portfolio returns
Authors: Sven Lehmann - University of Rostock (Germany)
Rainer Alexander Schuessler - University of Muenster (Germany) [presenting]
Mark Trede - University of Muenster (Germany)
Abstract: A simple framework is proposed where sparse equal-weighted subsets enhance long-term portfolio performance when return signals are weak and noisy. Building on a prior study, which shows that even randomly selected, equal-weighted small portfolios can outperform the market when rebalanced, this insight is translated into a systematic signal-based selection that amplifies the effect. The design combines simple predictive signals, such as short-term momentum with L0-based subset selection and equal weighting, to contain estimation error and shift the bias-variance trade-off in favor of more robust realized long-term compound returns.
