Fluke of stochastic volatility versus GARCH inevitability or which model creates better forecasts

The paper proposes the thorough investigation of in-sample and out-of-sample performance of five GARCH and two stochastic volatility models, estimated on the Russian financial data. The data includes prices of Aeroflot and Gazprom stocks and Ruble against US dollar exchange rates. In our analysis we use probability integral transform for in-sample comparison and Mincer-Zarnowitz regression along with classical forecast performance measures for out-of-sample comparison. Studying both the explanatory and the forecasting power of the considered models we came to the conclusion that stochastic volatility models perform equally or in some cases better than GARCH models.