In this paper, by enhancing the penalty coefficient, the general square loss criterion was modified into a novel criterion to more precisely identify risks and returns. Then, under this criterion, the ideal asset allocation for pension fund participants was investigated considering wealth management fees before retirement. Then, the Hamilton-Jacobi-Bellman (HJB) equation was formulated through the dynamic programming approach, and both the optimal investment strategy and minimum loss function were determined using calculus methods. Finally, how important parameters affect the initial optimal investment strategy and minimum loss function was analyzed, their economic implications were explained, the rationality of the model was validated, and several recommendations for management were provided.