DOI: 10.17093/alphanumeric.1901302 ISSN: 2148-2225

Global technology company stock price forecasting using long short-term memory (LSTM) architecture under macro financial variables

Aynur İncekırık
Show PDF Cite
This study aims to analyze the price dynamics of technology stocks such as Apple (USA), Samsung Electronics (South Korea), Xiaomi (China), Sony (Japan), LG Electronics (South Korea), and Nokia (Finland) using deep learning models. The analysis also includes exchange rates such as the US Dollar Index (DXY) along with the Chinese Yuan (USD/CNY), Japanese Yen (USD/JPY), and South Korean Won (USD/KRW) against the US Dollar. The study uses daily opening and closing prices for the period 09.06.2018–09.02.2026; relationships between variables were examined using Pearson correlation analysis. Stock prices, dollar index, and exchange rate data were obtained from the "Yahoo Finance" website. The results show a strong positive correlation between Apple and Sony, Apple and Samsung, and Samsung and Sony. In contrast, Xiaomi has a moderate positive correlation with Apple and Samsung, while the relationship between Apple and LG and Apple and Nokia remains weak. These findings reveal that sector-based common factors are influential in pricing. In the forecasting process conducted with three different layered LSTM models, the data was divided into training and test sets while maintaining chronological integrity; the models were evaluated using RMSE, MAE, and MAPE metrics. The results show that the performance of LSTM models varies by variable; while the single-layer LSTM architecture produced lower errors in most stock and currency series, two- or three-layer structures proved superior in some series. Overall, the study demonstrates that deep learning approaches are effective in modeling the price behavior of global technology companies and offer a strategic decision support tool in sustainable finance.

More from our Archive