A Gold Price Prediction Model Based on Economic Indicators Using Temporal Convolution and Attention Mechanism
Xueer Niu, Wanjiang Wang, Xinchen Leng, Xuena Guo · 2024
This study developed a gold price prediction model named TCDAB by analyzing key economic indicators such as Inflation Expectations, Federal Funds Rate, the US Dollar Index, and historical Gold Prices. The model integrates a temporal convolution strategy to observe long-term trends, a multi-head attention mechanism to enhance sensitivity to critical information, and a BIGRU model to identify gold price characteristics from both forward and backward aspects. This comprehensive approach aims to assist investors in understanding market dynamics more effectively, making crucial decisions, and maximizing returns. Experimental results show that the TCDAB model performs well in predicting gold prices for the next 5, 10, and 30 days, providing important decision support for investors.