Empirical studies on dynamic trading strategies with autoregressive assets

• Main Author: Goh, Chian Yi
• Other Authors: PUN Chi Seng
• Format: Final Year Project
• Language: English
• Published: Nanyang Technological University 2021
• Subjects: Science::Mathematics::Applied mathematics::Operational research; Business::Finance::Financial management
• Online Access: https://hdl.handle.net/10356/146086
• Institution: Nanyang Technological University

In this paper, the out-of-sample performances of the sample-based multi-period dynamic mean-variance models and global minimum-variance models are evaluated under both time-consistent and time-inconsistent (precommitment) setting. Across the eight empirical datasets we apply, the time-consistent strategies outperform the time-inconsistent strategies as the latter always suggests extreme portfolio weights which are less applicable and less robust in practical. Among time-consistent strategies, the performance of the time-consistent global minimum-variance model is superior to other strategies in minimizing the potential risk, even in the presence of estimation error. Whereas the time-consistent mean-variance model attains a recommendable risk-adjusted performance even though it does not outperform the benchmark, equally weighted portfolio consistently, in term of Sharpe ratios. Besides, for the time-consistent mean-variance strategy, the advancement extended to the autoregressive assets does suggest improvement in the performance, in term of risk-adjusted metrics including Sharpe ratios and CEQ returns. Enhancement through adjusting the desired required return may lead to a better achievement in balancing risk and return of asset allocation. This paper contributes to the evaluation of the practical use of dynamic trading strategies and hence the best choice among them, which is previously questionable.