Based on the Markowitz portfolio model, this paper considered the liquidity of the risk assets, set the minimum expected return rate of investors, and changed the inequality constraint to semi-gradient fuzzy number. Then used the penalty factor to adjust the objective function, taking the maximum liquidity and minimum risk of portfolio as the objectives, we established the Multi-objective portfolio optimization model. Using the weekly return rate and weekly turnover rate data of eight stocks that are typical in several industries, we can get the Pareto optimal solution set of risk assets investment proportion by fuzzy genetic algorithm. Investors can accord to the personal attitude toward return, risk, liquidity to choose better investment combination proportion of risk assets.