Understanding Gold and Dollar Price Movements: A Sentiment-Based GARCH-MIDAS Approach

Generally, gold prices exhibit an inverse relationship with the US Dollar Index (DXY). However, a notable deviation occurred in 2020 and 2022 when both markets displayed synchronized upward trends, challenging conventional financial theories. To explore this anomaly, this study proposes an innovative framework by integrating sentiment indicators into the GARCH-MIDAS model, thereby constructing a GAERCH-MIDAS-Sentiment hybrid model. Leveraging high-frequency market data and textual sentiment indices extracted from news and social media, the research quantifies behavioral factors influencing gold’s safe-haven attributes during macroeconomic shocks. Empirical results demonstrate that the augmented model reduces out-of-sample prediction errors by 18.7% compared to traditional volatility models, while significantly improving the explainability of tail risks during crisis periods. Specifically, sentiment-driven herding effects amplified by pandemic uncertainties and geopolitical tensions were identified as critical channels driving the gold-DXY correlation shift. This methodology advances precious metal market analysis by bridging behavioral finance with macroeconometric modeling, offering institutional investors a dynamic tool for portfolio hedging strategies under structural break scenarios. The findings further underscore the necessity of incorporating nonlinear sentiment dynamics in commodity pricing models amid escalating global market interconnectedness.