Scholarlink

Stock price prediction is of the utmost importance for financial analysis as accurate forecasting can have a major impact on an investment strategy and risk management. In this work, we investigate the performance of LSTM, RNNs, and GRUs for stock price prediction by providing a comparative analysis of these models in predicting eBay Inc.’s stock price. In addition, we propose the use of recent optimizer algorithms like NAdam and AdamW to improve the models’ performance. The RNN improved from an R2 of 91.4% to 93.4% when using NAdam instead of Adam, showing the effectiveness of our method. We additionally show that GRU with the Adam optimizer algorithm consistently outperforms other models across all metrics.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Deep Learning Methods for Stock Prediction: Comparative Analysis of LSTM, RNN, GRU and Optimizer Performance

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Quoc Hung Nguyen,
  • Ngoc Thien Nguyen,
  • Pham Ha An Bui,
  • Ngoc Uyen Nhi Pham,
  • Phuong Anh Nguyen

摘要

Stock price prediction is of the utmost importance for financial analysis as accurate forecasting can have a major impact on an investment strategy and risk management. In this work, we investigate the performance of LSTM, RNNs, and GRUs for stock price prediction by providing a comparative analysis of these models in predicting eBay Inc.’s stock price. In addition, we propose the use of recent optimizer algorithms like NAdam and AdamW to improve the models’ performance. The RNN improved from an R2 of 91.4% to 93.4% when using NAdam instead of Adam, showing the effectiveness of our method. We additionally show that GRU with the Adam optimizer algorithm consistently outperforms other models across all metrics.