The value function of the cumulative prospect theory is introduced into portfolio selection, and the traditional integer subscripts hesitant fuzzy linguistic set is improved from the perspective of behavioral finance. The improved variable subscripts hesitant fuzzy linguistic set can not only reflect investors' uncertainty about financial products, but also describe investors' psychological perception of gains or losses when giving evaluations. From the angle of single-investor multi-criteria evaluation, different stocks are quantitatively evaluated by the hesitant fuzzy language multi-criteria portfolio evaluation system based on variable subscripts. Maximum returns and minimum risks hesitant fuzzy linguistic portfolio models are proposed for the three types of investors: aggressive, moderate and conservative. The optimal solutions of hesitant fuzzy language portfolio are obtained by solving the equivalent nonlinear models. Finally, numerical simulations verify the effectiveness of the proposed models and methods.