Bitcoin price prediction with other commodity prices as exogenous inputs using machine learning techniques

This study addresses a gap in the literature by predicting Bitcoin prices using commodity prices as exogenous variables, a focus previously unexplored. Bitcoin, often referred to as digital gold, has gained significant attention from investors worldwide due to its resilience during financial distress. Prior research primarily utilised macroeconomic indicators, technical indicators, or combinations of commodity prices and macroeconomic factors. However, our study exclusively examines the predictive power of commodity prices - gold, silver, copper, crude oil, and iron ore - on Bitcoin's price, employing machine learning techniques such as random forest, K-nearest neighbours, decision tree, extreme gradient boost, and linear regression. All models showed strong performance when evaluated against 11 error metrics. The findings underscore a robust correlation between Bitcoin and these commodities, with the machine learning models achieving high accuracy in forecasting Bitcoin price fluctuations. These insights hold valuable implications for investors and the broader financial research community.