Predict Gold Price Trend Based on ARIMA Model

Abstract

As a financial product, gold is one of the more important spot and futures trading products in the commodity market. Based on the time series model, the gold price can be fitted and predicted, in order to explore the law of gold price changes. It has positive implications for investors and government managers. This article selects the Prime Day price in 2018 as the research object. Combined with domestic and foreign research content on financial time series. First, through the time series diagram test and the unit root test, it is obtained that the gold daily price series is a cycle-free and non-stationary series. Therefore, the time series needs to be differentiated. Second, a new stationary sequence is obtained by making second-order differences. Third, after the time sequence diagram test, ADF test, and white noise test, the sequence is a non-white noise sequence. Comparing the AIC values of multiple time series models, the most ideal model for the series should be the ARIMA (2,2,2) model. The significance test of the model shows that the fitted model is significantly effective. And the significance test of the model parameters is also passed. Then make predictions with this model. Comparing the predicted value with the future real gold price, it is found that the predicted value is close to the real value. This is a good reference for the country to formulate relevant economic policies.