Hybrid Machine Learning Architecture for Stock Market Price Prediction: Integrating Statistical Time-Series Models with Deep Learning and Ensemble Methods
Sridhanush Varma1, R. Swejan Rao2, P. Ravi Teja3, M. Shiva4, B. Venkata Ramana5
1Sridhanush Varma, Student, Department of Data Science, Holy Mary Institute of Technology and Science, Hyderabad (Telangana), India.
2R. Swejan Rao, Student, Department of Data Science, Holy Mary Institute of Technology and Science, Hyderabad (Telangana), India.
3P. Ravi Teja, Student, Department of Data Science, Holy Mary Institute of Technology and Science, Hyderabad (Telangana), India.
4M. Shiva, Student, Department of Data Science, Holy Mary Institute of Technology and Science, Hyderabad (Telangana), India.
5Dr. B. Venkata Ramana, Associate Professor, Department of Data Science, Holy Mary Institute of Technology and Science, Hyderabad (Telangana), India.
Manuscript received on 06 February 2026 | Revised Manuscript received on 09 March 2026 | Manuscript Accepted on 15 March 2026 | Manuscript published on 30 March 2026 | PP: 13-22 | Volume-14 Issue-4, March 2026 | Retrieval Number: 100.1/ijese.F833514060326 | DOI: 10.35940/ijese.F8335.14040326
Open Access | Editorial and Publishing Policies | Cite | Zenodo | OJS | Indexing and Abstracting
© The Authors. Blue Eyes Intelligence Engineering and Sciences Publication (BEIESP). This is an open-access article under the CC-BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/)
Abstract: Stock prediction is hard. Prices are noisy, non stationary, and nonlinear. We built a hybrid system that combines statistical models (ARIMA, GARCH), deep learning (LSTM, GRU), and Random Forests via Ridge regression meta-learning. The meta-learner uses 5-fold time-series cross-validation to adaptively weight models. Testing across 20 stocks from Technology, Finance, Healthcare, Consumer, and Industrial sectors, we achieved 87.74% average RMSE improvement over individual models. Directional accuracy ranged from 42.45% to 5.87%. Boeing (BA) showed 95.43% RMSE improvement with 5.87% directional accuracy; U.S. Bancorp (USB) hit 94.31% RMSE improvement. Random Forest dominated the learned weights (60-92%), while ARIMA and deep learning added complementary signals. Walk-forward validation with 252-day rolling windows ensured that we tested on truly unseen data, not on retrofitted history.
Keywords: Stock Price Prediction, Hybrid Machine Learning, ARIMA, GARCH, LSTM, GRU, Random Forest, Ensemble Learning, Meta-Learning, Time-Series Forecasting
Scope of the Article: Computer Science and Engineering