Efficiency analysis in the cryptocurrency market and return prediction

This study aims to determine the current volatility structures of Bitcoin, Ethereum, Binance Coin, Solana, and Ripple, which are among the cryptocurrencies with the highest market value, and to determine the market efficiencies by determining whether they exhibit long memory features and to make return predictions with the PATSOS, which is a hybrid machine learning method. In the study, daily data of cryptocurrencies between the years 2014-2022 were used. According to the results of the study, the existence of long memory in volatility has been determined in Bitcoin, Binance Coin, Solana, and Ripple return series. Therefore, it has been determined that the Efficient Market Hypothesis in volatility is not valid for the return series of the cryptocurrencies except Ethereum. As a result of the tests, the most suitable and sufficient models for volatility prediction in return series were found as HYGARCH (1, d, 1) for Bitcoin, IGARCH (1, 1) for Ethereum, FIGARCH (1, d, 0) for Binance Coin, and FIGARCH (1, d, 1) and FIAPARCH (1, d, 1) models for Solana and Ripple. In addition, lower RMSE, MSE, and MAE error values with higher percentages of correct forecast were obtained with the PATSOS method compared to the ANFIS method for all cryptocurrencies.