Abstract

The growing complexity of financial markets requires intelligent systems that can process vast amounts of multimodal data while offering transparent decision rationales. Traditional algorithmic trading models often focus solely on quantitative numerical indicators, neglecting market sentiment and macroeconomic textual signals. In this paper, we introduce IntelliTradeAI, a tri-signal fusion framework designed for explainable AI-powered financial market prediction. IntelliTradeAI combines historical quantitative price trends, sentiment signals extracted from financial news and social media via transformer models, and macroeconomic indicators into a unified attention-based representation. To ensure transparency, the framework incorporates SHAP-based feature importance attribution, illuminating the primary drivers behind predictive trading signals. Experimental results demonstrate that IntelliTradeAI outperforms conventional baseline models in predictive accuracy while providing interpretable insights essential for risk management and algorithmic trading.