In the intricate financial markets, understanding and forecasting price movements of Exchange-Traded Funds (ETFs) price movements demand innovative approaches. This research pioneers a methodology that synergizes network analysis with spatiotemporal graph embeddings, notably Node2Vec, to bolster the transparency and interpretability of financial prediction models. We transform financial markets' multifaceted and intertwined structures into a more digestible, low-dimensional space, thereby amplifying the efficacy of AI-driven predictions. Central to our approach is deploying the SHAP Explainable AI (xAI) technique, shedding light on the decision-making intricacies of our tree-based models. With a suite of six distinct tree-based models at our disposal, our predictions achieve superior accuracy and champion the cause of transparency and explainability. Ultimately, our approach, anchored by Node2Vec, paves a pathway for stakeholders to navigate the often murky waters of AI models, granting them invaluable insights into the undercurrents driving financial market dynamics.