Volatility dynamics of crypto-currencies' returns: Evidence from asymmetric and long memory GARCH models

The objective of this paper is to select the most optimum model or set of models useful for modeling sixteen of the most popular crypto-currencies associated volatility. Five GARCH models, with different error distributions, are fitted to each of these crypto-currencies. The most effectively fit model or superior set of models is then selected through maximizing the likelihood and minimizing the AIC and BIC information criteria. The reached results prove that the majority of crypto-currencies turn out to be rather effectively modulated via the TGARCH with double exponential distribution. Indeed, the attained findings report an asymmetric effect whereby volatility turns out to increase rather by response to positive shocks than by response to negative shocks, implying an asymmetric effect that differs from that generally observed in stock markets. The increase in volatility, as emanating in response to positive shocks may well have its justification in the uninformed investors' undertaken herding strategies.