Portfolio optimization remains a critical task in finance, aiming to attain an ideal equilibrium between the level of risk and the corresponding return. Many approaches often rely on metrics such as the Sharpe and Value at Risk (VaR) ratios to evaluate portfolio performance and risk exposure. This paper applies the above indicators with fuzzy numbers to incorporate uncertain factors into the model, thereby helping the model reflect reality more closely. Using these metrics in portfolio selection models poses challenges due to their objective functions’ nonlinear and non-convex nature. Instead of using the existing heuristic approach, we propose a new approach based on pseudo-convex programming. Theoretically, we prove the pseudo-convexity of the fuzzified objective functions, thereby showing that the equivalent problem is pseudoconvex. Taking advantage of the elegant properties of pseudo-convex programming, we propose an efficient algorithm to tackle the multi-objective fuzzy portfolio optimization problem. Experimentally, we demonstrate the effectiveness of our approach through empirical analyses using real-world financial data and its superiority over multi-objective evolutionary algorithms such as NSGA-III and AGE-MOEA-II.