Does Anything Beat a GARCH(1,1)? Evidence from Crypto Markets

We leverage the emerging cryptocurrency markets as an independent testing ground to evaluate the hypothesis that the GARCH(1,1) model serves as an effective and generalizable volatility model due to its simplicity and parsimony. Our investigation extends to five variations of the standard GARCH model across six cryptocurrency markets: Bitcoin (BTC), Ethereum (ETH), Litecoin (LTC), Binance Coin (BNB), Ripple (XRP), and Chainlink (LNK). Consistent with findings from conventional financial markets, our analysis endorses the GARCH(1,1) model as a suitable prior for cryptocurrency volatility modeling, demonstrating superior performance in four out of the six markets examined. Specifically for Bitcoin, incorporating an E-GARCH model to account for asymmetric volatility responses to positive and negative market movements offers significant modeling advantages, despite the increased complexity and associated model risk.