Scholarlink

The rise in investment is driven by increased public awareness of innovative financial management strategies. The stock market remains a leading investment choice, attracting significant public interest. Many investors choose companies without thorough analysis, often leading to financial losses due to declining stock values. This element is largely as a result of stock prices being highly unpredictable and experiences constant changes and direction in flow. Deep learning techniques have been employed by researchers to predict stock prices by examining data, including historical stock price trends from prior years. Therefore, the objective of this study is to forecast stock prices using the long short-term memory (LSTM) method, specifically focusing on the five top LQ45 companies in Indonesia. The empirical findings from hyperparameter tuning demonstrate that LSTM is capable of predicting the stock price with an average RMSE value below the stock price standard deviation. Furthermore, the R2 score likewise indicates strong performance, with an average value of 0.92 across all companies.

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

Stock Price Forecasting Model for the Five Leading LQ45 Companies in Indonesia Using a Long Short-Term Memory Approach

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Lili Ayu Wulandhari,
  • Natasha Hartanti Winata,
  • Nathania Christy Nugraha,
  • Aditya Kurniawan,
  • Syarifah Diana Permai,
  • Rindang Widuri,
  • Zirawani Baharum

摘要

The rise in investment is driven by increased public awareness of innovative financial management strategies. The stock market remains a leading investment choice, attracting significant public interest. Many investors choose companies without thorough analysis, often leading to financial losses due to declining stock values. This element is largely as a result of stock prices being highly unpredictable and experiences constant changes and direction in flow. Deep learning techniques have been employed by researchers to predict stock prices by examining data, including historical stock price trends from prior years. Therefore, the objective of this study is to forecast stock prices using the long short-term memory (LSTM) method, specifically focusing on the five top LQ45 companies in Indonesia. The empirical findings from hyperparameter tuning demonstrate that LSTM is capable of predicting the stock price with an average RMSE value below the stock price standard deviation. Furthermore, the R2 score likewise indicates strong performance, with an average value of 0.92 across all companies.