Volume 14 | Issue 5
Volume 14 | Issue 5
Volume 14 | Issue 5
Volume 14 | Issue 5
Volume 14 | Issue 5
This empirical research study is centred around the utilization of neural networks to forecast gold price volatility with an outstandingly low estimation error of just 0.036. The predictive analysis involves the inclusion of crucial independent variables, such as Crude Oil and silver prices, non-farm payroll data, the Dollar Index, CPI, and the Retail Index. These variables are drawn from a ten-year historical dataset, with great care taken in collecting secondary data from diverse sources, including the Chicago Mercantile Exchange website and official U.S. government platforms that provide payroll insights. The gathered data is then input into a neural network model characterized by a specific architecture featuring two hidden layers, each composed of two neurons. This model is meticulously refined over 25 iterative steps to achieve optimal performance. Notably, the model impressively demonstrates significant success by producing minimal root mean square values when subjected to testing against a dedicated dataset. A remarkable feature of this study is the consistency in error across the predicted variables, which serves to bolster the model's predictive reliability.