Automation of the Individualized Investing Strategy for an Investment Advisor Established by a Semi-Markov Regime-Switching Model

The purpose is to establish an automated investing strategy which can imitate an advisor’s behaviour in financial market. In view of the above needs, we review previous studies of Markov regime-switching model whose duration is geometrically distributed, propose a semi-Markov regime-switching model whose duration has a general distribution. By extending the state space of the semi-Markov chain, the model is transformed to a Markov regime-switching model. As the full information of the semi-Markov regime-switching model is available in the issue, we propose a divide-and-conquer and computationally tractable algorithm to estimate parameters. Experiments with empirical datasets show that the automated investing strategy based on estimated parameters behaves like the investment advisor. For an investment advisor, the automated investing strategy can help the advisor to avoid boring routines, and evaluate the advisor’s advice thoroughly.