Explainable AI for Financial Market Prediction Using Time-Series Data

Explainable AI for Financial Market Prediction Using Time-Series Data

Mohammad Amin Eftekhari1

1) Department of computer science and electrical engineering, Shahid Beheshti University Tehran,Iran

IOI Code : AAGF-AAGBKZ
Publication : 9th International Conference on Applied Researches in Science & Engineering (carseconf.com/9th)
Abstract :
    The rapid growth of Artificial Intelligence (AI) has significantly transformed financial market prediction, particularly through advanced time-series data analysis. Despite their predictive accuracy, black-box models such as Long Short-Term Memory (LSTM) networks and Convolutional Neural Networks (CNNs) face criticism for their lack of interpretability, which limits trust and adoption in financial decision-making. To address this challenge, the concept of Explainable Artificial Intelligence (XAI) has emerged as a promising solution, providing insights into model behavior while maintaining high predictive power. This paper explores the integration of XAI techniques, including SHapley Additive exPlanations (SHAP) and Local Interpretable Model-Agnostic Explanations (LIME), to enhance the transparency of AI-driven financial forecasting. Using historical stock market time-series data, we evaluate the performance of deep learning models combined with XAI frameworks. The experimental results indicate that while LSTM and CNN achieve strong forecasting accuracy, the incorporation of SHAP and LIME enables a deeper understanding of feature contributions, such as volatility, trading volume, and macroeconomic indicators. Our findings demonstrate that explainability not only improves investor confidence but also supports regulatory compliance, risk management, and practical deployment of AI models in financial markets. This study highlights the necessity of balancing accuracy and interpretability for robust and trustworthy financial prediction systems.

Keywords : explainable ai time series data financial market prediction long short term memory convolutional neural networks shapley additive explanations interpretable model agnostic explanations