Application of ARIMA and ARIMA-GARCH models in predicting stock prices for British American Tobacco

Abstract

Forecasting stock prices is important for investors as it allows them to make the right decisions. Trends in historical data can be effectively captured by time series models such as ARIMA (Autoregressive Integrated Moving Average), but they are not able to capture the inherent volatility of financial markets. This volatility can lead to significant forecasting errors, impacting investment decisions. Generalized Autoregressive Conditional Heteroskedasticity (GARCH) will also be used to address this limitation by modeling volatility. This study investigates the potential of a hybrid ARIMA-GARCH model for predicting stock prices using historical data. The hybrid model successfully captured volatility clustering, significantly enhancing forecasting accuracy. A comparison based on Theil’s U-statistic showed a lower value for the ARIMA-GARCH model compared to ARIMA alone. This robust hybrid framework offers valuable insights into the price dynamics and volatility trends of BAT’s stock, benefiting investors and financial analysts.