Due to the fast development of computational intelligence, algorithmic modeling has become an indispensable method used for financial market analysis. This paper proposes an investor sentiment index by using genetic algorithm to explore its effect on post-earnings-announcement dynamics in Chinese A-share market. Based on the data of Shanghai Stock Exchange, a multi-dimensional investor sentiment index is built using MDA genetic dual-machine optimization process with consideration of non-linear relationship among sentiment indicators. Our empirical findings indicate that investor sentiment has a clear impact on stock return volatility for the period around earnings announcements. Particularly, positive earnings surprises generate stronger and faster price reactions in high-sentiment firms than that of low-sentiment ones. In low sentiment periods, on the other hand, negative earnings announcements elicit more sustained and severe price declines. These results reveal a one-sided asymmetry sentiment effect between optimistic and pessimistic investors, and provide evidence that the behavioral factors still play roles in stock price reaction after earnings announcement.