An artificial bee colony algorithm for uncertain portfolio selection.

Area of Science:

• Finance
• Operations Research
• Data Science

Background:

• Traditional portfolio selection often relies on historical data, which may not accurately reflect future security returns.
• Incorporating transaction costs and diversification is crucial for practical portfolio management.
• Expert evaluations offer an alternative to historical data for estimating security returns, especially in uncertain market conditions.

Purpose of the Study:

• To develop a novel portfolio selection model that utilizes expert evaluations and accounts for transaction costs and diversification.
• To introduce a mean-variance-entropy framework for measuring investment return, risk, and portfolio diversification.
• To design an efficient algorithm for solving the proposed portfolio optimization problem.

Main Methods:

• Utilizing uncertain variables to model security returns based on expert evaluations.
• Developing a mean-variance-entropy model where uncertain mean, uncertain variance, and entropy represent return, risk, and diversification, respectively.
• Designing a modified artificial bee colony (ABC) algorithm to solve the optimization model.

Main Results:

• The proposed mean-variance-entropy model effectively integrates expert evaluations, transaction costs, and diversification degree.
• The modified ABC algorithm demonstrates effectiveness in solving the complex portfolio selection problem.
• Numerical examples validate the practical applicability and efficiency of the developed model and algorithm.

Conclusions:

• Expert evaluations combined with uncertain variables provide a robust approach to portfolio selection.
• The mean-variance-entropy model offers a comprehensive framework for portfolio optimization under uncertainty.
• The developed ABC algorithm is a viable tool for addressing real-world portfolio adjustment challenges.