ПОСТРОЕНИЕ И ВЕРИФИКАЦИЯ МНОГОФАКТОРНОЙ ЭКОНОМЕТРИЧЕСКОЙ МОДЕЛИ ДЛЯ ПРОГНОЗИРОВАНИЯ ВОЛАТИЛЬНОСТИ НА РЫНКЕ КРИПТОАКТИВОВ С ИСПОЛЬЗОВАНИЕМ GARCH-МОДЕЛЕЙ

Abstract:
The highest volatility is the main characteristic of the crypto asset market, which creates an urgent need for accurate forecasting models for risk management and algorithmic trading tasks. This article builds and verifies a multifactorial econometric model for volatility forecasting based on an extended KARCH approach. The paper uses data on the returns of BTC and ETH for the period from January 1, 2020, to December 31, 2023. A comparative analysis of the GARCH, EGARCH and GJR-GARCH specifications has been carried out. The analysis showed that the EGARCH asymmetric model (1,1) most accurately describes the observed data. To increase the adequacy of the model, it was improved by adding external variables to the equation describing the change in variance.: VIX volatility index and trade turnover. When tested on a deferred sample, it was found that the extended EGARCH-X model, which takes into account external factors, demonstrates statistically significantly superior predictive ability compared to the basic models, which is confirmed by using the Diebold-Mariano test. The conclusions drawn indicate the importance of simultaneously considering such volatility characteristics as clustering and asymmetry, as well as the influence of external market factors, to achieve more accurate forecasts in the cryptocurrency market.