Gold Price Forecasting in Uncertain Times: Integrating Sentiment Analysis and External Indices

Since 2020, financial markets have experienced unprecedented volatility, making the traditional models for predicting gold prices significantly challenging. These models often overlook the nuanced impacts of real-time sentiment, economic indicators, and related metals like silver. This study proposes a novel framework to combine gold prices with other relevant features to improve predictive performance in the precious metal market. Further, we employ three advanced recent deep learning models, LSTM, N-HiTS, and TSMixer, and evaluate their performance with various metrics using the different combinations of the proposed features. The results demonstrate that incorporating the proposed additional features, such as silver prices and sentiment analysis, significantly improves forecasting accuracy, amidst the volatile and uncertain market conditions spanning from 2020 to 2024. The best-performing models outperform traditional models, such as ARIMA. Our findings highlight the importance of feature selection and combination in predictive modeling and demonstrate the potential of multi-feature models in capturing complex market dynamics. The insights from this study can inform investment decisions and risk management strategies in the precious metal market, and future research can build on our approach to explore additional features and models.