Copyright © 2025 Authors retain the copyright of this article. This article is an open access article distributed under the Creative Commons Attribution License which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

@article{180251,
        author = {SUBHENDU AKHULI and PROF. BISWADEV GOSWAMI and PROF. ABHIJIT CHOWDHURY},
        title = {Stock Price Prediction, using Recurrent Neural Networks and Optimized Deep Learning Models},
        journal = {International Journal of Innovative Research in Technology},
        year = {2025},
        volume = {12},
        number = {1},
        pages = {897-908},
        issn = {2349-6002},
        url = {https://ijirt.org/article?manuscript=180251},
        abstract = {This project presents a comprehensive study on the use of Long Short-Term Memory (LSTM) networks, a specialized class of Recurrent Neural Networks (RNNs) within the domain of deep learning, for time series forecasting of stock prices. The research specifically focuses on Tata Steel Limited, a key stock within the Indian equity market, selected for its historical volatility and market relevance. Accurate prediction of stock prices is a highly challenging task due to the non-linear, non-stationary, and noisy nature of financial time series data. Traditional statistical methods often fall short in capturing these complex temporal patterns, which underscores the need for more sophisticated, data-driven approaches.
The core objective of this study is to investigate the efficacy of bidirectional LSTM architectures in modelling and predicting future stock price movements based on historical data. To this end, the model is trained on Open, High, Low, and Close (OHLC) price features retrieved from Yahoo Finance, with optional inclusion of technical indicators such as Moving Averages (MA), Relative Strength Index (RSI), and Bollinger Bands to enhance pattern recognition. The Adam optimizer is utilized for gradient-based optimization of the model parameters due to its computational efficiency and adaptive learning capabilities.
The model’s predictive performance is quantitatively evaluated using Root Mean Squared Error (RMSE), a robust metric for measuring the deviation between predicted and actual values. Additionally, the research examines the influence of hyperparameters, particularly the number of training epochs, on the model’s convergence behavior and forecasting accuracy. Experimental results suggest that LSTM networks, when properly tuned and supplemented with relevant indicators, can effectively learn from historical patterns and outperform baseline models in terms of predictive precision.
Beyond model performance, the project also explores the practical challenges associated with implementing deep learning models in financial forecasting, including overfitting, data pre-processing complexities, and interpretability issues. Overall, this study contributes valuable insights to the growing field of machine learning in finance, illustrating how RNN-based models like LSTM can serve as powerful tools for stock market analysis, investment decision support, and algorithmic trading.},
        keywords = {Deep Learning, Time Series Forecasting, Stock Price Prediction, Recurrent Neural Network (RNN), Long Short-Term Memory (LSTM), OHLC Data, Adam Optimizer, Root Mean Squared Error (RMSE), Technical Indicators, Financial Market Analysis, Hyperparameter Tuning, Model Convergence, Algorithmic Trading},
        month = {June},
        }