Scholarlink

This paper explores utilizing Machine Learning algorithms, specifically LSTM, for stock price prediction. The dynamic nature of the financial markets presents a significant challenge for investors and analysts in making informed decisions. Fundamental and technical analysis which are traditional methods of stock price prediction, have been supplemented by advanced ML algorithms because they can analyze historical data and recognize underlying patterns. This research aims to bridge the theoretical and practical aspects of stock price forecasting by utilizing LSTM model to process 45 months of historical data (from March 2020 to January 2024) of Nifty 50. The dataset contained low, adjusted close, high, volume traded, and open prices obtained from Yahoo Finance. The study adopts an exploratory research design, using the Weka software tool for model deployment, and evaluation was done with and without the application of technical indicators – EMA, ROC & RSI to the data, separately and combined. The models are assessed based on their prediction accuracy, employing metrics such as MAE, RMSE, RAE and RRSE. Experimental results indicate that LSTM outperforms traditional ML models (SVM, RF, and Decision Tree) in predicting stock prices, demonstrating lower error values and higher correlation coefficients in all market volatile conditions. This study promotes the field by providing empirical evidence of the efficacy of LSTM model in stock price prediction, offering valuable insights for investors, financial analysts, and policymakers.

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

Predicting Stock Price Movements with Recurrent Neural Networks: An LSTM-Based Approach

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • V. Srividya,
  • M. Chandramohan,
  • Rathimala Kannan,
  • R. Sujatha

摘要

This paper explores utilizing Machine Learning algorithms, specifically LSTM, for stock price prediction. The dynamic nature of the financial markets presents a significant challenge for investors and analysts in making informed decisions. Fundamental and technical analysis which are traditional methods of stock price prediction, have been supplemented by advanced ML algorithms because they can analyze historical data and recognize underlying patterns. This research aims to bridge the theoretical and practical aspects of stock price forecasting by utilizing LSTM model to process 45 months of historical data (from March 2020 to January 2024) of Nifty 50. The dataset contained low, adjusted close, high, volume traded, and open prices obtained from Yahoo Finance. The study adopts an exploratory research design, using the Weka software tool for model deployment, and evaluation was done with and without the application of technical indicators – EMA, ROC & RSI to the data, separately and combined. The models are assessed based on their prediction accuracy, employing metrics such as MAE, RMSE, RAE and RRSE. Experimental results indicate that LSTM outperforms traditional ML models (SVM, RF, and Decision Tree) in predicting stock prices, demonstrating lower error values and higher correlation coefficients in all market volatile conditions. This study promotes the field by providing empirical evidence of the efficacy of LSTM model in stock price prediction, offering valuable insights for investors, financial analysts, and policymakers.