Forecasting stock price using the ARIMA-GARCH model

The purpose of this article is to build a model for forecasting financial time series. The article considers an econometric approach involving the construction of an autoregressive model - an integrated moving average (ARIMA), as well as a generalized model with autoregressive conditional heteroscedasticity (GARCH). As an enhancement of the predictive power of the model, it is proposed to use a combination of the above-mentioned models. A mathematical description of these predictive models is given. Model building is implemented in the Python software environment with connected libraries pandas, numpy, statsmodels, matplotlib. As input data sets, the daily values of Alrosa stock quotes were imported from 06/02/2014 to 11/12/2019. The results of the study show that the combination of ARIMA-GARCH models has high predictive accuracy and can be used to create short-term forecasts, and it is also concluded that the transition to more adaptive models that take into account external factors.