Behavioural perspectives in forex portfolio value analysis

This paper combines Cumulative Prospect Theory (CPT) and the Grey Clustering Algorithm (GCA) to guide the optimization of forex portfolio selection. The United States Dollar (USD) against a universe of 84 other currencies was used for portfolio value analysis using the Differential Evolution Algorithm. A total of six portfolios were constructed of which two were based on the CPT and the remaining on the GCA. The optimisation results of all constructed portfolios show that the GC-based portfolios outperformed the CPT-based portfolios. Specifically, GC Portfolio 4, comprising assets with higher CPT values in GC 1, emerged as the best-performing portfolio with a Sharpe ratio of 0.8497, significantly surpassing the highest Sharpe ratio among CPT-based portfolios (0.0206 for CPT Portfolio 2), further reinforcing the superiority of GCA in portfolio optimisation. The inclusion of the behavioural proxy in portfolio construction has a significant impact on adding value to investors' portfolios. Future research could explore refining asset selection by integrating machine learning techniques such as K-means clustering or reinforcement learning to enhance portfolio robustness. This research paper explores the integration of Cumulative Prospect Theory (CPT) and Grey Clustering Algorithm (GCA) for forex portfolio optimization. The study is significant as it introduces a behavioural finance framework that accounts for investors' cognitive biases and decision-making anomalies, which traditional portfolio models often overlook. By evaluating the performance of portfolios composed of 84 currencies, the research demonstrates the superiority of GCA-based portfolios over those based on CPT, showcasing a higher Sharpe ratio and thus a better risk-return profile. This work makes a meaningful contribution to the literature on portfolio optimization, offering new insights into behavioural finance and demonstrating the practical value of integrating behavioural proxies into asset selection. Additionally, it suggests that advanced clustering techniques like GCA can enhance the reliability of portfolio analysis, making the study relevant for investment managers and financial analysts seeking to refine their strategies in volatile and uncertain market conditions. Future research could further expand on these findings by incorporating machine learning models to improve asset selection and portfolio robustness.