Forecasting Stock Market Crashes using Deep Learning

  1. Olivia Chatterjee
  2. Alok Kole
  3. Indranil SenGupta
  4. Arnab Majumdar

Reviewed by

Read the full article See related articles

Discuss this preprint

Start a discussion What are Sciety discussions?

Listed in

Log in to save this article

Abstract

This paper addresses the important issue of predicting stock market crashes, which can cause major problems due to large drops and long recovery times. We present a new method that uses a mix of machine learning and deep learning to improve prediction accuracy and provide useful insights in financial forecasting. Our approach includes Support Vector Machines (SVMs), Deep Reinforcement Learning (DRL), Long Short-Term Memory (LSTM) networks, Convolutional Neural Networks (CNNs), Generative Adversarial Networks (GANs), and Deep Deterministic Policy Gradient (DDPG) models. An important part of our work is the development of a DRL model specifically designed to be more resilient during market downturns. This model can help reduce financial losses during crises and also provide significant returns for investors. The experimental results strongly support the robustness of our DRL model in unpredictable market conditions. Performance checks, which use R2-Score, Mean Squared Error (MSE), Mean Absolute Error (MAE), and Receiver Operating Characteristic (ROC) curves, confirm that our methods outperform models like Graph Neural Networks (GNNs) and Transformers. This DRL model achieves a high true positive rate for detecting crash signals while keeping the false-alarm rate low. Its strength comes from the reinforcement learning agent’s ability to learn decision-making strategies that take misclassification costs into account within a simulated trading environment.

Article activity feed

  1. PREreview

    This Zenodo record is a permanently preserved version of a Structured PREreview. You can view the complete PREreview at https://prereview.org/reviews/17636334.

    Does the introduction explain the objective of the research presented in the preprint? Yes
    Are the methods well-suited for this research? Somewhat appropriate
    Are the conclusions supported by the data? Somewhat supported
    Are the data presentations, including visualizations, well-suited to represent the data? Somewhat appropriate and clear
    How clearly do the authors discuss, explain, and interpret their findings and potential next steps for the research? Somewhat clearly
    Is the preprint likely to advance academic knowledge? Moderately likely
    Would it benefit from language editing? Yes
    Would you recommend this preprint to others? Yes, but it needs to be improved
    Is it ready for attention from an editor, publisher or broader audience? No, it needs a major revision

    Competing interests

    The author declares that they have no competing interests.

    Use of Artificial Intelligence (AI)

    The author declares that they did not use generative AI to come up with new ideas for their review.

  2. Version published to 10.20944/preprints202510.1781.v1