Optimizing Precious Metal Price Forecasting with Hybrid Deep Learning Models: An Operational Research Perspective

In the current study, Wolfram Mathematica is employed to study the application of seven deep neural network architectures in predicting precious metal prices: LSTM, CNN, MLP, MLP-CNN, MLP-LSTM, LSTM-CNN, and LSTM-CNN-LSTM. We will utilize a daily price series over ten years for gold, silver, and platinum. Model performances were evaluated using RMSE, MAPE, and RMSE metrics. Results illustrate the strong performance of all neural network models as such techniques will capture long-term dependencies for better decision-making on the commodity markets. As compared to the individual neural networks, the hybrid neural network show superior performance. The LSTM-CNN-MLP model is a very dependable and robust solution for precious metals forecasting.This will clearly shows that NN plays a major role while predicting the time series data. These results contribute to the wealth of literature in operational research with improved predictive accuracy in financial systems and portfolio optimization strategies.