The rapid adoption of artificial intelligence (AI) has made it essential to forecasting of AI-related assets volatility. This study employs explainable artificial intelligence methods, specifically decision tree, extra tree, random forest, gradient boosting models to forecast future volatility in AI assets. The study results reveal that the gradient boosting model achieves the highest predictive directional accuracy 80% to 90% for volatility. Furthermore, SHAP analysis identifies key predictors, which including technical indicators such as the moving average, exponential moving average, and momentum, which effectively capture short-term market sentiment, while macroeconomic variables reflect broader economic conditions. This research bridges a gap in the literature by integrating explainable machine learning with time-series forecasting of AI asset volatility, offering insights into which model and variables most effectively forecast volatility in both blockchain-based and traditional AI asset markets, which is essential for effective portfolio optimisation and risk management.
Ali et al. (Thu,) studied this question.