Abstract

Cryptocurrency markets are highly volatile, driven by rapidly shifting factors such as social media sentiment, trading volume, and macroeconomic signals. While deep learning (DL) models offer strong predictive performance, their black-box nature limits their adoption in decision-critical environments like crypto investing (Yang et al., 2023). Existing research primarily focuses on improving prediction accuracy but neglects the interpretability of model outputs (Basu et al., 2023). As a result, traders and institutional investors lack visibility into the rationale behind forecasts, reducing trust and limiting regulatory alignment. There is a need for frameworks that combine predictive accuracy with transparent explanations of what drives market behavior. This study draws on bounded rationality theory and behavioral finance, emphasizing that investor decisions depend on simplified cues and trust in model transparency. The integration of SHapley Additive exPlanations (SHAP) into DL predictions can enhance interpretability and guide more informed decision-making. We propose a conceptual framework that integrates DL models— such as LSTM and BERT—with SHAP to predict and explain cryptocurrency price trends(Bauer et al., 2023). The research explores: Which market indicators (e.g., volume, sentiment, volatility) most influence model predictions? How can SHAP enhance transparency in DL-based trading tools? The study is currently conceptual. We aim to build a DL pipeline using historical crypto price data and social sentiment, followed by SHAP-based feature importance analysis. Future validation will involve simulation-based back testing and stakeholder interviews. This research contributes to the growing literature on explainable AI in finance by demonstrating how SHAP can bridge the gap between accuracy and transparency in crypto prediction. Practically, it informs the development of interpretable AI tools for institutional investors, traders, and regulators navigating high-risk digital asset environments.

Comments

tpp1467

Download

Share

COinS