This paper regards the monthly announced reports of fund performance evaluation as the target and proposes a skyline-based filtering method for selecting good funds, which may be ignored by the other method. Based on these funds, we further analyze the movement of TAIEX and build a system for the analysis of mutual funds and the prediction of market trend. The experiment results show that our method is superior to the commonly used rules in the subsequent performance of the selected funds. The trend prediction achieves about 60% in accuracy, and our method for skyline computation can also speed up above 20%.