Accurate Conditional Variance Models for Predicting Asymmetric Volatility in Cryptocurrency Markets

This study includes tests on the Generalized Autoregressive Conditional Heteroscedasticity (GARCH) model and its derivatives to conduct complex and detailed volatility analysis for the 5 highest-volume cryptocurrencies traded in September 2023. The tests have been conducted with Python, R, and Eviews software and analyses have been compared in terms of consistency and accuracy of the results across multiple software and programming languagse. In the testing process, observation of the volatility has been assessed by some variables such as skewness, kurtosis, and log-likelihood values, and these variables have been taken into consideration for testing. Tests such as Jarque-Bera and Augmented Dickey-Fuller (ADF) have been applied during the process to verify model correctness. The EGARCH, GJR-GARCH, and TGARCH models have been more effective in detecting volatility and market shocks in the relevant cryptocurrencies as a result of the tests conducted in the volatility analysis.