The selection of an appropriate threshold is one of the main concerns about the POT (Peaks Over Threshold ) model, which is widely used in many fields, such as in insurance and environmental analysis. In this article, a new method which is based on the MSE criterion and two-sample K-S tests is proposed to determine the optimal threshold in the GPD model. Different from the traditional subjective methods, the proposed method can determine the threshold by quantification, and is easy to be implemented by a computer program. An empirical study using the daily returns of S&P 500 index is also presented. It's shown that the proposed method can well capture the optimal threshold.