EURO 2025 Leeds
Abstract Submission

3107. Implementing non-dominated sorting into asset preselection within portfolio problem

Invited abstract in session MB-9: Data Science in Insurance and Finance: New perspectives and Applications, stream OR in Finance and Insurance .

Monday, 10:30-12:00
Room: Clarendon SR 2.01

Authors (first author is the speaker)

1. Tomáš Tichý
Department of Finance, Faculty of Economics, VSB-Technical University Ostrava
2. David Neděla
Finance, VSB-TU Ostrava
3. Sergio Ortobelli Lozza
University of Bergamo

Abstract

By monitoring the financial markets, we can discover a huge number of potential investments, such as various stocks. Notwithstanding, it is very complicated and time-consuming to analyze all stocks and then use all of them in optimization according to the preferred portfolio strategy. Hence, for portfolio managers, it can be useful to focus on just a specific selection identified by a proper filtering process. This paper proposes an efficient approach for stock preselection based on the multidimensional nondominated sorting of selected statistics. In contrast to previous research, we design an innovative application based on statistics obtained from approximated return series using nonparametric regression and principal component analysis (PCA). Additionally, we study the impact on mean–variance and recently proposed complex mean–trend risk large-scale portfolio selection strategies. In particular, this process examines the efficient frontier of portfolios on the basis of various return and risk perspectives. In the empirical application to US stock market data, we present an ex-post and ex-ante analysis with results for 40 portfolio strategies. The results obtained strongly indicate that for most risk-averse investors,
the mean–trend risk strategies with preselection outperform the same strategies without preselection, as well as the mean–variance strategies.

Keywords

Status: accepted

Back to the list of papers