metaheuristic

Diversifying investment portfolios by incorporating a variety of assets is a well-established strategy for mitigating risk and enhancing returns. Traditionally, mathematical models for portfolio optimization have primarily focused on stock investments within the capital market. However, this study extends the scope of portfolio optimization to encompass both project and stock investments. This is a critical advancement as investors increasingly grapple with allocating budgets across these two asset types simultaneously. Therefore, this paper proposes a novel mixed portfolio optimization model that uses the Mean-SemiVariance-SemiEntropy approach. By incorporating project investments alongside traditional stocks, the proposed model offers more efficient portfolios that can lead to improved return/risk ratios for investors seeking to optimize their overall financial strategy.

Design/methodology/approach:

 An attempt has been made to bridge the gap between the distinct spaces of projects and stocks to facilitate their joint analysis. Subsequently, a Mean-SemiVariance-SemiEntropy approach has been employed to develop a model within a probabilistic framework. For validating this model, a numerical experiment involving three projects and five stocks from the capital market has been tackled, considering the preferences of an investor. Finally, the optimization problem has been solved using three metaheuristic algorithms: Genetic Algorithm (GA), Imperialist Competitive Algorithm (ICA), and Gray Wolfs Optimization (GWO).

The results obtained by solving the model using the above-mentioned metaheuristic algorithms demonstrated that despite the high speed of the GWO algorithm, the solutions provided by the GWO algorithm were not satisfactory compared to the GA and ICA algorithms. On the other hand, the acceptable speed with nondominated solutions was the advantage of the ICA algorithm over the GA algorithm. The evaluation of various performance metrics also revealed that the ICA algorithm outperformed the GA and GWO algorithms in this problem. Also, the inclusion of semi-entropy as a risk assessment metric led to an improvement in the return on the investment portfolios.

Research limitations/implications: 

Incorporating investor constraints and preferences, such as cardinality and boundary constraints, into the model forms an NP-hard problem. Consequently, exact solution methods are replaced by non-exact methods, such as metaheuristic algorithms. Given the diversity of project contracts, this study concentrated solely on projects with cost-plus contracts, where the entire project or a portion can be selected for partnership. Similar to the Markowitz model, the projects' returns such as stocks' returns were assumed to be normally distributed.Practical implications: This study significantly enhanced diversification, increased potential returns, and reduced risk for investors by introducing a novel mixed project-and-stock portfolio optimization model. The proposed approach can be implemented by a wide range of investors and managers of investment units in various organizations, bringing a new perspective to investment management.

Social implications: 

The far-reaching implications of this study extend beyond the realm of investment management, permeating social, economic, and political areas. The innovative mixed project-and-stock portfolio problem has the potential to positively transform society using fostering innovation, stimulating economic growth, and enhancing financial knowledge. This paper can foster economic growth and job creation by providing new investment opportunities and increasing investment in productive ventures. In summary, this study has taken a significant step towards improving social and economic well-being by introducing an innovative model for resolving investment challenges.

Originality/value:

 The innovation and strength of this research lies in incorporating projects as a new asset class into the traditional portfolio model. This goes beyond simply adding a new asset to an investment portfolio, as the nature of the projects introduces new complexities to the portfolio management process. For this purpose, this study employs a probabilistic approach based on historical data. In addition, the simultaneous use of two risk measures, i.e., semi-variance and semi-entropy, significantly improves the performance of the model by focusing on different risk aspects. This provides a more comprehensive picture of the risks associated with the portfolio and helps investors make more informed and wise decisions.