Comparative Analysis of ARIMA, Multiple Linear Regression, and LSTM Models for Stock Price Prediction: Evidence from Starbucks and Luckin Coffee

Fujia Zhang

doi:10.54254/2754-1169/2025.BL28705

Advances in Economics, Management and Political SciencesOpen access

Comparative Analysis of ARIMA, Multiple Linear Regression, and LSTM Models for Stock Price Prediction: Evidence from Starbucks and Luckin Coffee

Research Article

Open Access

Comparative Analysis of ARIMA, Multiple Linear Regression, and LSTM Models for Stock Price Prediction: Evidence from Starbucks and Luckin Coffee

Fujia Zhang ^1*

¹ Anhui University

^*Corresponding author: R22314054@stu.ahu.edu.cn

Published on 28 October 2025

AEMPS Vol.233

ISSN (Print): 2754-1177

ISSN (Online): 2754-1169

ISBN (Print): 978-1-80590-485-4

ISBN (Online): 978-1-80590-486-1

Download Cover

Abstract

This paper compares three forecasting approaches-autoregressive integrated moving average (ARIMA), multivariate linear regression (MLR), and long short-term memory networks (LSTM)-for daily stock prices of Starbucks (SBUX) and Luckin Coffee (LKNCY). Trading calendars are aligned across tickers, missing observations are forward-filled, and technical indicators are engineered (log returns and lags, SMA/EMA, RSI, and volume change). A chronological split (80% train, 20% test) prevents look-ahead bias. Performance is evaluated using RMSE, MAE, and MAPE; Diebold-Mariano (DM) tests assess pairwise differences in forecast errors. On LKNCY, MLR attains the lowest error (RMSE 0.891; MAPE 2.74%), outperforming a baseline ARIMA (RMSE 7.155) and a univariate LSTM trained on prices with min-max scaling (RMSE 6.849). Results on SBUX show the same ranking. Diebold-Mariano tests show no statistically significant difference between LSTM and ARIMA forecast errors (SBUX: p=0.52; LKNCY: p=0.70). Diagnostics further indicate that, during downtrends, the price-target LSTM drifts toward the lower bound of the training range, a pattern consistent with sensitivity to scaling and distributional shift. To mitigate this issue, a robustness variant is introduced that predicts next-period log-returns using z-score standardization with a multi-layer LSTM. Taken together, the results emphasize the short-horizon strength of simple linear baselines and the central role of target choice, scaling, and evaluation protocol in financial time-series forecasting.

Keywords:

stock price forecasting, ARIMA, multiple linear regression, LSTM

View PDF

References

[1]. R. J. Hyndman and G. Athanasopoulos, Forecasting: Principles and Practice, 3rd ed. OTexts, 2021. Available at: https: //otexts.com/fpp3/

[2]. R. S. Tsay, Analysis of Financial Time Series, 3rd ed. Hoboken, NJ: Wiley, 2010. Available at: https: //www.wiley.com/en-us/Analysis%2Bof%2BFinancial%2BTime%2BSeries%2C%2B3rd%2BEdition-p-9780470414354

[3]. G. E. P. Box, G. M. Jenkins, G. C. Reinsel, and G. M. Ljung, Time Series Analysis: Forecasting and Control, 5th ed. Hoboken, NJ: Wiley, 2015. Available at: https: //www.wiley.com/en-us/Time%2BSeries%2BAnalysis%3A%2BForecasting%2Band%2BControl%2C%2B5th%2BEdition-p-9781118675021

[4]. A. Pankratz, Forecasting with Dynamic Regression Models. New York: Wiley, 1991. doi: 10.1002/9781118150528. Available at: https: //onlinelibrary.wiley.com/doi/book/10.1002/9781118150528

[5]. S. Hochreiter and J. Schmidhuber, “Long short-term memory, ” Neural Computation, vol. 9, no. 8, pp. 1735–1780, 1997. doi: 10.1162/neco.1997.9.8.1735. Available at: https: //direct.mit.edu/neco/article/9/8/1735/6109/Long-Short-Term-Memory

[6]. T. Fischer and C. Krauss, “Deep learning with long short-term memory networks for financial market predictions, ” European Journal of Operational Research, vol. 270, no. 2, pp. 654–669, 2018. doi: 10.1016/j.ejor.2017.11.054. Publisher page: https: //www.sciencedirect.com/science/article/pii/S0377221717310652 ; Open-access preprint: https: //econpapers.repec.org/RePEc: eee: ejores: v: 270: y: 2018: i: 2: p: 654-669

[7]. O. B. Sezer, M. U. Gudelek, and A. M. Ozbayoglu, “Financial time series forecasting with deep learning: A systematic literature review (2005–2019), ” Applied Soft Computing, vol. 90, art. 106181, 2020. doi: 10.1016/j.asoc.2020.106181. Open-access preprint: https: //arxiv.org/abs/1911.13288

[8]. F. X. Diebold and R. S. Mariano, “Comparing predictive accuracy, ” Journal of Business & Economic Statistics, vol. 13, no. 3, pp. 253–263, 1995. doi: 10.1080/07350015.1995.10524599. Publisher page: https: //www.tandfonline.com/doi/abs/10.1080/07350015.1995.10524599 ; OA version: https: //www.ssc.wisc.edu/~bhansen/718/DieboldMariano1995.pdf

[9]. S. Makridakis, E. Spiliotis, and V. Assimakopoulos, “The M4 Competition: Results, findings, conclusion and way forward, ” International Journal of Forecasting, vol. 34, no. 4, pp. 802–808, 2018. doi: 10.1016/j.ijforecast.2018.06.001. Publisher page: https: //www.sciencedirect.com/science/article/pii/S0169207018300785 ; OA post-print (example): https: //researchportal.bath.ac.uk/files/192035719/IJF_2019_M4_Editorial_post_print_.pdf

[10]. R. J. Hyndman and Y. Khandakar, “Automatic time series forecasting: The forecast package for R, ” Journal of Statistical Software, vol. 27, no. 3, pp. 1–22, 2008. doi: 10.18637/jss.v027.i03. Available at: https: //www.jstatsoft.org/v27/i03

[11]. N. T. Nhan, “Top 10 Fast Food Giants: Stock Price Dataset (2024), ” Kaggle Datasets, 2024. License: CC0 Public Domain. Files used: SBUX.csv, LKNCY.csv. Available at: https: //www.kaggle.com/datasets/nguyentiennhan/stock-prices-of-the-10-largest-fast-food-companiesAccessed: Sept. 10, 2025.

References

[1]. R. J. Hyndman and G. Athanasopoulos, Forecasting: Principles and Practice, 3rd ed. OTexts, 2021. Available at: https: //otexts.com/fpp3/

[4]. A. Pankratz, Forecasting with Dynamic Regression Models. New York: Wiley, 1991. doi: 10.1002/9781118150528. Available at: https: //onlinelibrary.wiley.com/doi/book/10.1002/9781118150528

Cite this article

Zhang,F. (2025). Comparative Analysis of ARIMA, Multiple Linear Regression, and LSTM Models for Stock Price Prediction: Evidence from Starbucks and Luckin Coffee. Advances in Economics, Management and Political Sciences,233,83-90.

Data availability

The datasets used and/or analyzed during the current study will be available from the authors upon reasonable request.

About volume

Volume title: Proceedings of ICFTBA 2025 Symposium: Data-Driven Decision Making in Business and Economics

ISBN: 978-1-80590-485-4(Print) / 978-1-80590-486-1(Online)

Editor: Lukášak Varti

Conference website: https://2025.icftba.org/Bratislava.html

Conference date: 12 December 2025

Series: Advances in Economics, Management and Political Sciences

Volume number: Vol.233

ISSN: 2754-1169(Print) / 2754-1177(Online)

© 2024 by the author(s). Licensee EWA Publishing, Oxford, UK. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license. Authors who publish this series agree to the following terms:

1. Authors retain copyright and grant the series right of first publication with the work simultaneously licensed under a Creative Commons Attribution License that allows others to share the work with an acknowledgment of the work's authorship and initial publication in this series.

2. Authors are able to enter into separate, additional contractual arrangements for the non-exclusive distribution of the series's published version of the work (e.g., post it to an institutional repository or publish it in a book), with an acknowledgment of its initial publication in this series.

3. Authors are permitted and encouraged to post their work online (e.g., in institutional repositories or on their website) prior to and during the submission process, as it can lead to productive exchanges, as well as earlier and greater citation of published work (See Open access policy for details).