Predicting Market Behavior with Artificial Neural Networks: Gold Price as an Example

Abstract

Due to the highly nonlinear and random nature of the financial time series, forecasting the price of a product of interest is a very challenging task. Artificial neural networks excel at connecting diverse data sets, which has significant potential for commercial operations. The current study investigates the possibility of applying machine learning techniques to forecast the price of a target product by using information from other stock indexes. The gold price was the selected product while the stock indexes were S&P500, NASDAQ, DAX 40, Dow-jones, Nikkei, and oil prices. The data used in the study covered the previous 10 years. The simulations showed that neural networks can be used for making accurate predictions for the price of gold for the next 50 days.