نوع مقاله : پژوهشی
تازه های تحقیق
عنوان مقاله English
نویسندگان English
In recent years, statistical time series models—particularly the ARIMA model—have been widely used as effective tools for stock price forecasting. Despite its acceptable accuracy, the model’s sensitivity to the number of historical data points remains a major limitation in practical applications. This study aims to enhance prediction accuracy and stability by proposing a new model called RARIMA (Reset Auto Regressive Integrated Moving Average), developed based on the ARIMA framework. The proposed model employs a Fast-Learning Reset approach to reduce the effect of the selected historical data length. Stock price data from 18 companies across four industries—steel, petrochemical, banking, and automotive—were analyzed over the period April 2018 to January 2024. To evaluate model performance, the Mean Relative Error (MRE) and the TOPSIS ranking method were used to compare short-term and long-term forecasting accuracy. The results indicate that the RARIMA model improves prediction accuracy by up to 75% compared to the traditional ARIMA model, while significantly reducing its sensitivity to the size of historical data. Accordingly, the proposed model can serve as a reliable and efficient tool for financial analysts and decision-makers in the capital market.
کلیدواژهها English
Copyright © Mansoureh Naderipour, Hadis Al-Sadat Hosseini, Mohammad-Bagher Jamali
License
This article is released under the Creative Commons Attribution (CC BY 4.0) license. Anyone is free to copy, share, translate, and adapt this article for any purpose, whether commercial or non-commercial, as long as proper citation is given to the authors and original publication.
