摘要：This paper attempts to extend the literature on modeling goldprices by studying a unique set of variables using data from 2000 to2015. I construct a family of forecasting models which I then refinebased on the significance of multivariate linear regression results.The best model carries a predictive capacity of around 30%, and themost powerful indicators from the variables used are the USD-GBPexchange rate and the S&P 500 stock market index.
