Light buoy is a navigation aid sign to guide the navigation of ships, which plays an important role in ensuring the navigation safety of vessels. In order to predict the light buoy's offset position, and provide accurate buoy's position for shipping, a mathematical model of light buoy's offset prediction is established using multiplicative seasonal ARIMA model based to the seasonal time series characteristics of buoy's position data. Taking the No. 1 light buoy in Meizhou Bay as an example, the model is used to predict the offset position of the light buoy and compare with the actual position data. The results show that the multiplicative seasonal ARIMA model performance well, and the average absolute percentage errors of azimuth and shift distance are 1.79% and 7.37% respectively. This provides a new idea for the prediction of light buoy offset position.