A Machine Learning Approach to Gold Price Prediction Using Financial Indicators

This study presents a machine learning framework for predicting gold prices by integrating diverse financial indicators, including the NASDAQ-100 index (^NDX), Bitcoin (BTC-USD), and gold futures (GC=F).Using daily high prices from February 2020 to May 2024, the approach incorporates robust preprocessing techniques such as the Box-Cox transformation and Principal Component Analysis (PCA) to address skewness, kurtosis, and multicollinearity, to reduce dimensionality while retaining 96.37% of the variance.A Genetic Algorithm-optimized Multi-Layer Perceptron (MLP) regression model achieved high predictive accuracy with an R² score of 0.98, an RMSE of 23.48 USD, and an MAE of 17.38 USD.Permutation importance analysis highlighted PC1 and PC2 as the most significant predictors, collectively capturing over 96% of the dataset's variance.The results emphasize the effectiveness of integrating stock indices, cryptocurrencies, and traditional financial variables for gold price prediction.This research offers practical applications for investors and policymakers by offering insights into market trends, enhancing decisionmaking, and bridging traditional and emerging markets in financial forecasting.