Volume 14, Number 1, June 2024
| An Analysis of Taiwan Stock Market Price Forecasting During International Panic Events: A Study on Bear Market Periods |
Kuang-Hsun Shih 6, Ying-Che Ting 7*, Jui-Cheng Hung 8, Yi-Hsien Wang 1*
Abstract
This study compares the effectiveness of ARIMA and LSTM models in predicting market behavior during global crises by analyzing historical panic events and assessing the impact of political, economic, and social disruptions on financial systems. The ARIMA model is employed to capture immediate, linear market volatility, while the LSTM approach examines extended, nonlinear responses to crisis situations. The findings highlight the strengths and limitations of each method under different market conditions, providing valuable insights for financial analysts and investors to enhance market interpretation and strategic decision-making during periods of uncertainty.
Keywords: Long Short-Term Memory, LSTM, ARIMA, Dual-Model Comparison
JEL Classification: C22, C45, G17
6 Doctorate Program in Intelligent Banking and Finance, CTBC Business School, Tainan, Taiwan.
* ( E-mail: wang12@ctbc.edu.tw)
7 Master Program of Business Administration in Practicing, Chinese Culture University,Taipei,Taiwan
8 Bachelor Program of Fashion Creative Industry & Branding Management, Chinese Culture University, Taipei,Taiwan