Accuracy Evaluation of Stock Index Forecasting Models

The study aims to create several forecasting models for Russian stock index (MICEX) based on stock indexes of other countries and macroeconomic indicators and subsequently evaluate the accuracy of corresponding models. The mentioned procedures are carried out in order to determine optimal forecasting models and the list of variables that generate the most accurate forecast. The analysis covers the period from January 2001 to March 2017 and two sub-periods: January 2001 – March 2009 and April 2008 – March 2017. The paper is based on application of the following models: ARIMA, VAR and VECM. The study investigates several key issues including inter alia accuracy comparison of selected models, efficiency evaluation of forecasting on the basis of stock indexes of other countries in the context of different time intervals analysis, comparison of forecasting accuracy of VAR and VECM specifications based on stock indexes of developed and emerging countries, as well as macroeconomic variables.