Comparative analysis of ARMA and ARIMA models for different forecasting horizons: choosing the optimal approach

The article examines the predictive power of historical data on stock quotes in the Russian stock market. The objective of the work is to evaluate the effectiveness of ARMA and ARIMA time series models for different stock quotes forecasting horizons. The study analyzed time series of closing quotes for Sberbank shares using various data sets and dividing the samples into testing and training. The results showed that in the current market conditions, historical quote values do not have sufficient predictive value, which limits the relevance of technical analysis. The findings of the study can be useful for traders and analysts when choosing approaches to forecasting the volatility of stock quotes. Further research is recommended into the use of more sophisticated tools that consider the influence of external and internal factors, quantitative and qualitative information.