Authors: Tushar Hingmire
Abstract: Gold has long been regarded as a safe-haven asset, yet its price is subject to intense volatility driven by a complex interplay of global economic conditions, currency fluctuations, and commodity market dynamics. Traditional forecasting methods often fail to capture these non-linear dependencies, motivating the development of data-driven approaches. This paper presents GoldMind AI, a machine learning framework designed to forecast gold prices using four key macro-financial indicators: the S&P 500 Index (SPX), the United States Oil Fund ETF (USO), the iShares Silver Trust ETF (SLV), and the EUR/USD currency pair exchange rate. Two supervised learning models — Linear Regression and Random Forest Regressor — are trained on historical financial data and evaluated using standard regression metrics including Root Mean Squared Error (RMSE), Mean Absolute Error (MAE), and the coefficient of determination (R²). The Random Forest model achieves an R² score of 0.92, RMSE of 1.07 USD, and MAE of 0.94 USD, significantly outperforming Linear Regression with a 23% reduction in error rates. The trained model is deployed as an interactive web application built with Streamlit, enabling real-time gold price forecasting from user-supplied market inputs. GoldMind AI demonstrates that ensemble machine learning methods can effectively capture complex market relationships, providing actionable insights for investors and financial analysts.
