NeuroStock: A Deep Learning Framework for Stock Market Forecasting with Web-Based Analytics

Yashika Hooda, Dr. Ekta Soni

Abstract: Stock market prediction is challenging due to its volatile, nonlinear nature. NeuroStock is a deep learning framework using Long Short-Term Memory (LSTM), Convolutional Neural Network (CNN), and Hybrid CNN-LSTM models to predict stock prices over 7-60 days. It processes real-time Yahoo Finance data and evaluates performance with metrics like RMSE, MAE, R², Directional Accuracy, and Precision. A Streamlit web app lets users select stocks, tweak settings, view predictions, and compare models. Tests on GOOG stock show the Hybrid CNN-LSTM model balances accuracy (RMSE: 9.218274, R²: 0.913406), while LSTM minimizes errors (RMSE: 4.413391). Unlike prior studies, NeuroStock excels in model comparison, trend prediction, and user-friendly deployment. It offers real-time insights for investors and analysts. Future updates, like sentiment analysis and additional data inputs, will boost accuracy, making NeuroStock a powerful tool for financial forecasting.

Keywords: Deep Learning, Stock Market Prediction, LSTM, CNN, Hybrid CNN-LSTM, Streamlit, Yahoo Finance, Time-Series Forecasting, Directional Metrics, Web Deployment

How to Cite?: Yashika Hooda, Dr. Ekta Soni, "NeuroStock: A Deep Learning Framework for Stock Market Forecasting with Web-Based Analytics", Volume 15 Issue 2, February 2026, International Journal of Science and Research (IJSR), Pages: 327-334, https://www.ijsr.net/getabstract.php?paperid=SR251028144252, DOI: https://dx.doi.org/10.21275/SR251028144252

Download Citation: APA | MLA | BibTeX | EndNote | RefMan