A Quest for Context-Specific Stock Price Prediction - A Comparison Between Time Series, Machine Learning and Deep Learning Models

Aug 5, 2026

Stock price forecasting supports both institutional and individual investors in making more informed buying, selling, and holding decisions amid complex market dynamics. This study analyzes Bombay Stock Exchange (BSE) data using time series, machine learning, and deep learning models across two datasets: one with COVID-19 stock prices and one without. Model comparisons identify performance strengths and limitations, showing time series models for short-term prediction, machine learning for stronger generalization, and deep learning for more accurate long-term forecasts. Understanding each model’s behavior helps investors and regulators optimize strategy and decision-making.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages24

Reading Time1 h 0 min

AccessFree to download

Abstract

Keywords and Abbreviations

Introduction

SN Computer Science (2025) 6:335 [https://doi.org/10.1007/s42979-025-03848-y](https://doi.org/10.1007/s42979-025-03848-y) A Quest for Context‑Specific Stock Price Prediction: A Comparison Between Time Series, Machine Learning and Deep Learning Models Mugdha Shailendra Kulkarni1 · S. Vijayakumar Bharathi1 · Arif Perdana2 · Divisha Kilari1 Received: 29 July 2023 / Accepted: 3 March 2025 © The Author(s) 2025 Abstract Understanding the complexities of buying, selling, and holding stocks is crucial for institutional and individual investors to make informed decisions. Despite their significance, many investors face challenges in this area. Accurate stock price forecasting is a vital tool that aids investors in making profitable decisions. This study evaluates stock trends and patterns with an in-depth analysis of the Bombay Stock Exchange (BSE) stock data. We utilized various techniques, including timeseries analysis, machine learning, and deep-learning models. This investigation spanned two distinct datasets: one with and one without COVID-19 stock price data. By comparing the outcomes, we seek to identify the most effective model for stock price prediction. Our findings indicate that each model has its strengths and limitations. Time series models accurately forecast short-term stock prices, whereas machine learning models demonstrate superior generalization capabilities. Deep learning models, however, stand out for their ability to predict long-term stock prices more accurately. Understanding each model's performance nuances is crucial for institutional and individual investors and regulators to optimize their strategies and decision-making processes. Keywords Stock price prediction · Machine learning · Deep learning · Time-series analysis · ARIMA · LSTM Abbreviations EIC Economic, industry, and company analysis GDP Gross domestic product ARIMA Auto-regressive integrated moving average SMA Simple moving average SVR Support vector regression RNN Recurrent neural networks LSTM Long short-term memory RFR Random forests regression XGBoost Extreme gradient boosting ACF Autocorrelation function * Arif Perdana [arif.perdana@monash.edu](arif.perdana@monash.edu) Mugdha Shailendra Kulkarni [mugdha@scit.edu](mugdha@scit.edu) S. Vijayakumar Bharathi [svkbharathi@scit.edu](svkbharathi@scit.edu) Divisha Kilari [kilari.divisha@associates.scit.edu](kilari.divisha@associates.scit.edu) 1 Symbiosis Centre for Information Technology, Symbiosis International (Deemed University), Pune, India 2 Monash University, Monash, Indonesia PACF Partial autocorrelation function AIC Akaike information criterion SBC Schwartz criterion BIC Bayesian information criterion MLP Multilayer perceptron ME Mean error RMSE Root mean square error MAE Mean absolute error MAPE Mean absolute percentage error MASE Mean absolute scaled error ACF1 Autocorrelation of errors at lag 1 RBF Radial basis function Introduction The stock market, a cornerstone of modern economies, has long captivated researchers and practitioners due to its complexity and far-reaching implications. At its core, stock price fluctuations are governed by fundamental principles of supply and demand [1] . This seemingly simple mechanism is, however, an intricate web of influencing factors that span the economic, industrial, and corporate domains [2–5] . Macroeconomic forces such as inflation rates, unemployment SN Computer Science figures, and gross domestic product have a significant influence on stock prices [6, 7] . However, the sensitivity of the market extends beyond quantifiable metrics. The power of the information, whether positive or negative, cannot be underestimated. Corporate events, from mergers and acquisitions to changes in governance structures, can cause dramatic shifts in share prices [8–10] . Indeed, unexpected negative news can be a particularly potent catalyst for market movements [11] . In this dynamic environme

FAQs

What data and evaluation context does the study use for stock price prediction?

The study analyzes Bombay Stock Exchange (BSE) stock data using two datasets: one including COVID-19 stock prices and one excluding them.

Which modeling families are compared in the research?

Time series analysis, machine learning models, and deep learning models are evaluated and compared.

How do the results differ across the model types?

Time series models are more accurate for short-term forecasts, machine learning models generalize better, and deep learning models more accurately predict long-term stock prices.

Aug 5, 2026

Upload by Evangeline