Predicting Commodity Prices in Futures Market Using Machine Learning

This study enhances predictive modelling of gold prices by employing advanced machine learning techniques, specifically Long Short-Term Memory (LSTM) networks and Convolutional Neural Networks (CNNs). Introducing an innovative Event Impact Score into the CNN-LSTM model, an improved forecasting accuracy was achieved compared with the models without this approach. This novel methodology that calculates an Event Impact Score based on economic events derived from an economic calendar dataset, quantifies the impact of significant economic indicators on commodity prices by assessing the associated price changes. Incorporating these scores, the CNN-LSTM model is adapted to accommodate the nuanced influence of external economic factors, offering a more refined analysis than conventional models.