Dynamic Market Behavior and Price Prediction in Cryptocurrency: An Analysis Based on Asymmetric Herding Effects and LSTM
Guangxi Cao,Meijun Ling,Jingwen Wei,Chen Chen
DOI: https://doi.org/10.1007/s10614-024-10676-4
IF: 1.741
2024-07-15
Computational Economics
Abstract:This study employs the cross-sectional absolute deviation model and Carhart pricing model to examine the existence and authenticity of various market sizes and liquidity levels within cryptocurrency markets. Additionally, we introduce a herding effect measurement index tailored for the cryptocurrency market and predict cryptocurrency prices by integrating the long short-term memory (LSTM) neural network model. Empirical results reveal the presence of both genuine and pseudo herding phenomena in cryptocurrency markets, with information acquisition asymmetry identified as a significant driver of herding behavior. Specifically, during market downturns in the overall market, only pseudo herding is observed in the upward market, whereas during periods of market prosperity, both genuine and pseudo herding are evident in the downward market. In markets of different sizes, herding is absent in cryptocurrency markets with small market value, while in large market value cryptocurrency markets, pseudo herding is not statistically significant. Genuine herding occurs in both upward and downward markets during non-downturn periods. Regarding cryptocurrency markets with different liquidity levels, herding behavior is not observed in markets with small trading volume. Conversely, in markets with large trading volume, pseudo herding is observed in both upward and downward markets during non-downturn periods, with genuine herding occurring in both markets during boom periods. Additionally, the LSTM model demonstrates superior capability in fitting the price trends of different cryptocurrencies, and considering the herding effect index significantly enhances the accuracy of cryptocurrency price prediction.
economics,mathematics, interdisciplinary applications,management