A Review of Stock Price Prediction Techniques using Machine Learning

Abstract

Stock price prediction is a challenging task due to the highly volatile and nonlinear nature of financial markets. This study explores various machine learning models, including traditional approaches such as Random Forest and XGBoost, deep learning models like Long Short-Term Memory (LSTM) networks and Gated Recurrent Units (GRU), and hybrid architectures incorporating external indicators. The models are trained on historical stock price data combined with external financial indicators to improve predictive accuracy. Performance is evaluated using key metrics such as RMSE, Mean Squared Error (MSE) and R-squared scores. Results indicate that researcher work with won generated dataset and compare in different traditional ML models-Random Forest / XGBoost, Support Vector Regression (SVR), etc. Deep Learning models- LSTM (Long Short-Term Memory), GRU (Gated Recurrent Unit), CNN (Convolutional Neural Networks) for Time-Series, Transformers (e.g., Informer, Time-Series Transformer) with external indicators, outperform traditional machine learning methods in capturing market trends.

Introduction

The equity market plays a crucial role in capital movement and economic growth. Predicting stock prices accurately is challenging due to their volatile and non-stationary nature, requiring advanced regression and classification models. This study focuses on structured stock data from Yahoo Finance and compares various machine learning (ML) and deep learning models for stock price prediction using evaluation metrics such as RMSE, MSE, and R².

A literature review highlights recent models like CEEMDAN-LSTM, SVR with MEMD, AdaBoost variants, and transformer-based approaches, each balancing predictive accuracy with computational complexity. Notably, LSTM-based models often outperform traditional ML models, especially when combined with feature-rich inputs.

The proposed system introduces a multi-input LSTM model that incorporates external indicators — technical indicators (RSI, MACD, moving averages), macroeconomic data, news sentiment, and market sentiment — alongside stock price and volume data. This approach aims to capture broader market influences, using LSTM’s gating mechanisms to better predict stock price movements by processing multiple data streams simultaneously.

Overall, integrating diverse data sources with advanced deep learning architectures holds promise for improving the accuracy and reliability of short-term stock price predictions.

Conclusion

The LSTM-based multi-input model enhances stock price prediction by combining historical stock prices and external indicators. This approach provides a more comprehensive view of market trends, leading to better predictive performance. As a technology in the eld of deep learning, neural network can solve non-linear problems well. LSTM neural network is optimized on traditional neural network and introduces the concept of \"gate\", which enhances the long-term memory ability of the model, which enhances its generalization ability. this paper constructs a stock price prediction model based on LSTM with Multi Input model and also chooses the use of singlefeature and multi-feature input models to seek better prediction results. Future scope will be Hyperparameter tuning, Feature selection, try (CNNLSTM) convolutional layers for better feature extraction.

References

[1] Yu Sun, SofianitaMutalib ,Nasiroh Omar, And LiweiTian, ”Novel Integrated Approach for Stock Prediction Based on Modal Decomposition Technology and Machine Learning”. VOL. 12, pp. 95209- 95222, July 2024, Digital Object Identifier 10.1109/ACCESS.2024.3425727 [2] Prediction of Stock Price Using Machine Learning- SwarnaPrabha Jena Department of ECE, Ankit Kumar Yadav Department of CSE, Dhiraj Gupta Department of CSE, Bijay Kumar Paikaray Faculty of Emerging Technologies,2023 IEEE 2nd International Conference on Industrial Electronics: Developments & Applications (ICIDeA) | 979-8-3503-8197-9/23/$31.00 ©2023 IEEE | DOI: 10.1109/ICIDeA59866.2023.10295232 [3] Sidra Mehtab, JaydipSen,” A Time Series Analysis-Based Stock Price Prediction Using Machine Learning and Deep Learning Models” April 2020,DOI:10.48550/arXiv.2004.11697,https://www.researchgate.net/publication/340938611 [4] Yi Li, Lei Chen, Cuiping Sun , Guoxu Liu , ChunleiChen,YonghuiZhang,”Accurate Stock Price Forecasting Based on Deep Learning and Hierarchical Frequency Decomposition” vol. 12, 11 April 2024, pp. 49878-49894, https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=10488422 [5] SaeedeAnbaeeFarimani, MajidVafaeiJahan And Amin MilaniFard,”An Adaptive Multimodal Learning Model for Financial Market Price Prediction”,DOI- 10.1109/ACCESS.2024.3441029, Vol. 12,10 September 2024, pp. 121846-121863. [6] XiaobinLu,Josiah Poon And MatloobKhushi (2024), \'Bridging the Gap between Machine and Human in Stock Prediction: Addressing Heterogeneity in Stock Market\', IEEE Access, 12, pp. 186171 - 186185. DOI: 10.1109/access.2024.3511613. [7] Mahdi PakdamanNaeini, HamidrezaTaremian ,HomaBaradaranHashemi,”Stock market value prediction using neural networks”, DOI:10.1109/CISIM.2010.5643675,IEEE Xplore, Conference: Computer Information Systems and Industrial Management Applications (CISIM), NOV. 2010 International Conference. [8] BaturGezici,EmreSefer,”DeepTransformer-Based Asset Price and Direction Prediction”,Jan.2024,IEEEAccess, PP:24164-24178,DOI:10.1109/ACCESS.2024.3358452. [9] KhorshedAlam , InjamamUlHaque , MdFahadMonir And Tarem Ahmed , “Enhancing Stock Market Prediction: A Robust LSTM-DNN Model Analysis on 26 Real-Life Datasets”, pp. 122757 – 122768, 29 July 2024 ,DOI: 10.1109/ACCESS.2024.3434524 [10] https://datasciencedojo.com/blog/how-do-i-make-an-lstm-model-with-multiple-inputs/, Time: 5:41 PM, 2-feb-2025 [11] A. A. Ariyo, A. O. Adewumi, and C. K. Ayo, ‘‘Stock price prediction using the ARIMA model,’’ in Proc. UKSim-AMSS 16th Int. Conf. Comput. Model. Simul., Mar. 2014, pp. 106–112. [12] W. Jiang, ‘‘Using the GARCH model to analyse and predict the different stock markets,’’ M.S. thesis, Uppsala Univ., Uppsala, Sweden, Dec. 2012, pp. 2–24. [13] Y. Funde and A. Damani, ‘‘Comparison of ARIMA and exponential smoothing models in prediction of stock prices,’’ J. Predict. Markets, vol. 17, no. 1, pp. 21–38, Jul. 2023. [14] B. B. Nair, V. P. Mohandas, and N. R. Sakthivel, ‘‘A decision tree-roughset hybrid system for stock market trend prediction,’’ Int. J. Comput. Appl.,vol. 6, no. 9, pp. 1–6, Sep. 2010. [15] Y. Lin, Y. Yan, J. Xu, Y. Liao, and F. Ma, ‘‘Forecasting stock index price using the CEEMDAN-LSTM model,’’ North Amer. J. Econ. Finance,vol. 57, Jul. 2021, Art. no. 101421. [16] Y. Huang, C. Deng, X. Zhang, and Y. Bao, ‘‘Forecasting of stock price index using support vector regression with multivariate empirical mode decomposition,’’ J. Syst. Inf. Technol., vol. 24, no. 2, pp. 75–95, Apr. 2022. [17] M. Ali, D. M. Khan, H. M. Alshanbari, and A. A.-A.-H. El-Bagoury,‘‘Prediction of complex stock market data using an improved hybrid EMDLSTM model,’’ Appl. Sci., vol. 13, no. 3, p. 1429, Jan. 2023. [18] L. Kang, X. Li, L. Wu, Y. Li, and X. Zhao, ‘‘Predicting stock closing price with stock network public opinion based on AdaBoost-IWOA-Elman model and CEEMDAN algorithm,’’ Surge, vol. 1, p. 4, Jan. 2022. [19] H. Ke, Z. Zuominyang, L. Qiumei, and L. Yin, ‘‘Predicting Chinese commodity futures price: An EEMD-Hurst-LSTM hybrid approach,’’ IEEE Access, vol. 11, pp. 14841–14858, 2023. [20] A. Akusta, ‘‘Exploring ceemdan decomposition for improved financial market forecasting: A case study on dow Jones index,’’ PamukkaleUniv.J. Social Sci. Inst., vol. 62, pp. 19–35, Apr. 2024 [21] Z. D. Ak?ehir and E. Kiliç, ‘‘A new denoising approach based on mode decomposition applied to the stock market time series: 2LE-CEEMDAN,’’PeerJ Comput. Sci., vol. 10, p. e1852, Feb. 2024. [22] I. Botunac, J. Bosna, and M. Mateti?, ‘‘Optimization of traditional stock market strategies using the LSTM hybrid approach,’’ Information, vol. 15,no. 3, p. 136, Feb. 2024. [23] J. P. Behura, S. D. Pande, and J. V. N. Ramesh, ‘‘Stock price prediction using multi-layered sequential LSTM,’’ EAI Endorsed Trans. Scalable Inf. Syst., vol. 11, no. 4, pp. 1–8, 2024. [24] M. Santoso, R. Sutjiadi, and R. Lim, ‘‘Indonesian stock prediction using support vector machine (SVM),’’ in Proc. MATEC Web Conf., vol. 164. Surabaya, Indonesia: Petra Christian Univ., 2018, pp. 1–5. [25] E. E. Peters, Fractal Market Analysis: Applying Chaos Theory to Investment and Economics, vol. 24. Hoboken, NJ, USA: Wiley, 1994.

Copyright

Copyright © 2025 Ashvini P. Wagh, Prof. Ashwini S. Gaikwad. This 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.