<p>Since Markowitz created modern portfolio theory, quantitative methods have gained increasing attention. However, the analytical formulas of constructed fuzzy portfolio models are often complex due to the influence of real-world constraints. To better align with actual investment situations, this paper integrates the irrational psychological factors of investors, assigning different risk perception coefficients based on investor types. Using possibility theory, we derive the possibility mean and lower semi-variance using these risk perception coefficients. To evaluate the performance of the security portfolio, three Data Envelopment Analysis-Banker Charnes Cooper models with different orientations are used to calculate the multi-stage efficiency and Malmquist index of the assets. These metrics serve as input and output indicators to assess the Decision-Making Unit. This paper constructs a mean-lower semi-variance Data Envelopment Analysis multi-stage efficiency-Malmquist index model, which comprehensively considers investors’ bounded rationality and realistic constraints. Utilizing actual market data from the Shanghai Stock Exchange, we verify the model’s rationality and effectiveness, demonstrating experimental results that closely align with real-life scenarios of personalized investor decision-making. The proposed model provides practical tools for practitioners and policymakers, enabling more informed and personalized investment decisions in complex financial environments.</p>