Improving stock market prediction with feature expansion and explainable AI

Abstract

The financial market is a complex, dynamic, and fast-paced environment where high-quality, precise, and timely information is highly valued. In this context, the ability to access and interpret financial data efficiently can mean the difference between success and failure in market operations. Price fluctuations and the inherent volatility of the stock market demand continuous analyses and constant updates. The objective of this research is to train deep learning models and utilize explainability techniques to filter indicators for both humans and machines. Through the use of advanced technologies, we aim not only to predict market movements but also to understand the underlying factors that influence these predictions. By doing so, we intend to provide valuable insights that can be used to improve investment strategies. We seek to evaluate the performance of these models in terms of execution time and prediction accuracy. To achieve this, various comparative experiments will be conducted, where different model architectures will be tested and adjusted to optimize their performance. Additionally, the explainability of the models will be a central focus, aiming to ensure that users can trust the generated predictions and understand the reasons behind each decision made by the algorithms. This study intends to contribute significantly to the field of computational finance, offering practical and theoretical solutions that can be applied by both individual traders and large financial institutions. Through the implementation of deep learning techniques and detailed analysis of their performance, we hope to identify best practices and propose improvements that can be adopted in future research and practical applications.