Year: 2025
Type: article
Abstract: The CSI 300 Index, a critical benchmark for China’s capital market, plays a pivotal role in investment decision-making and risk management. However, the non-stationary and volatile nature of financial time series poses challenges to traditional forecasting models. This study aims to evaluate the applicability of the ARIMA model in predicting short-term returns of the CSI 300 Index. Utilizing daily return data from 2010 to 2023, the research employs the Box-Jenkins methodology to construct an ARIMA model, involving stationarity tests (ADF and KPSS), differencing, and parameter optimization via ACF/PACF plots and information criteria (AIC/BIC). A rolling-window approach is adopted for out-of-sample forecasting. Results indicate that the ARIMA model achieves optimal performance in short-term (1-5 days) predictions, with a mean absolute error (MAE) of 0.45% for one-day-ahead forecasts. Residual diagnostics confirm the model’s adequacy (Ljung-Box test, p> 0.05). The findings demonstrate that ARIMA provides a cost-effective and efficient tool for high-frequency return predictions in China’s equity market, offering theoretical and practical insights for quantitative trading strategies. Limitations include the exclusion of macroeconomic factors, suggesting future integration with hybrid models (e.g., ARIMA-GARCH) for enhanced robustness.
Author: Linyue Zhang
Institution: Shanghai University
Language: English
Field: Decision Sciences
Domain: Social Sciences
Open Access status: green
OpenAlex ID: W7110916734
FWCI:
Cites:
Cited by:
Related to: