A novel grey wave forecasting method for predicting metal prices

Abstract

The evolution of metal prices shows severe fluctuations and irregular cycles which bring difficulties to accurate forecasting. This paper proposes a novel grey wave forecasting method with unequal-interval contour lines and contour time sequences filtrating to predict metal prices. In the proposed model unequal-interval contour lines are determined by the quantiles of data, which considers the intensity of data. Contour time sequences are filtrated based on autocorrelation characteristics of time series. Furthermore, we use monthly prices of two metals - aluminum and nickel-to assess the performance of our novel grey wave forecasting model with a multi-step-ahead prediction. The empirical analysis indicates the modified grey wave forecasting method is much better than basic grey wave forecasting method in terms of prediction accuracy and it can also achieve better forecasting results than ARMA and random walk.