An explainable deep learning approach for stock market trend prediction

Area of Science:

• * Computational Finance
• * Machine Learning
• * Artificial Intelligence

Background:

• * Stock market forecasting involves inherent risks and uncertainties, necessitating robust trend analysis for optimal investment and disinvestment strategies.
• * Accurate prediction of market trends is crucial for profit maximization and loss minimization.

Purpose of the Study:

• * To propose a deep learning model for predicting five distinct stock market trends: upward, downward, double top, rounded bottom, and rounded top.
• * To evaluate the proposed model's performance against common benchmarks like support vector machine, random forest, and logistic regression.
• * To enhance model interpretability using explainable AI (XAI) techniques.

Main Methods:

• * Development and implementation of a novel deep learning model for stock trend prediction.
• * Comparative analysis against support vector machine, random forest, and logistic regression using four diverse real-world datasets.
• * Application of SHAP (SHapley Additive exPlanations) and LIME (Local Interpretable Model-agnostic Explanations) for explainable AI.

Main Results:

• * The deep learning model achieved an average accuracy of 94.9%, significantly outperforming random forest (85.7%), support vector machine (60.07%), and logistic regression (52.45%).
• * The model demonstrated superior F1-score performance at 94.85%, compared to random forest (77.95%), support vector machine (21.02%), and logistic regression (46.23%).
• * SHAP analysis identified key influential features, enabling potential feature reduction while improving precision, recall, and F1-score, indicating a performance trade-off.

Conclusions:

• * The proposed deep learning model offers a powerful and interpretable solution for stock market trend forecasting.
• * The integration of XAI techniques provides valuable insights into prediction drivers, supporting investor decision-making.
• * The model's performance suggests practical applicability in financial risk management and strategic investment planning.