Geomagnetic low-point displacement method was proposed in the 20th century. It has been applied to forecast earthquakes in short term, with the base of characteristics of geomagnetic low-point in main land China. However, since this method analyses geomagnetic distortions before and after noon only, the accuracy of forecasting in practice is relatively poor. In this paper, we conducted a statistical analysis on 162 geomagnetic low-point displacement anomalies between 2008 and 2018. All the anomalies are computed by using daily variation spatial correlation method. Our results show that the accuracy of earthquake forecasting increases by 44%, from 39% to 56% when using mean correlation coefficient to determine the validities of anomalies. This method is proved to be effective and practical.