online gambling singapore online gambling singapore online slot malaysia online slot malaysia mega888 malaysia slot gacor live casino malaysia online betting malaysia mega888 mega888 mega888 mega888 mega888 mega888 mega888 mega888 mega888 虛擬貨幣價格形成與預測:情緒、總經、網絡與投資人結構:英文摘要



The purpose of the study is to analyze the determining factors of cryptocurrency returns. Using econometric models and machine learning, we extensively examine the fundamental, market microstructure, behavioral, investor structure (especially concentration), as well as technical dimensions of 6 major cryptocurrencies.

We find that compared with stock markets, the liquidity and ownership concentration of the crypto market is lower while volatility is higher. Crypto markets also show higher herding inclination, more trend following, and have lower modularity. From network analysis we find that typically cryptocurrencies have little connection with other markets (i.e., stock, gold, forex, oil) therefore they are good vehicles for risk diversification; however, during crises such as the 2020 pandemic and bearish markets, cryptocurrencies are connected to the other markets and may not be good hedges. Analyzing the crash and bubble periods for cryptocurrencies, we find the duration of both is about 10 days, but a bit longer and more intense for crashes. In terms of bubble, BSV has the biggest magnitude of price rises and the fastest speed of rise, while in terms of crashes DASH has the largest magnitude of drawdowns, also fastest speed of declines.

Market microstructure and investor behavior are the two main determinants of cryptocurrency returns. However, the significance of fundamental factors or concentration is not supported by empirical evidence. The higher liquidity, volatility, turnover or the stronger herding is, the more likely crypto returns will be in the top 25% or bottom 25%. Moreover, further analysis shows that, if investor attention is high in the high return state, chances of return reversal increases. If investor attention is low in the low return state, chances of return reverses to top 25% also increase, indicating the return reversal phenomena is strongly related with investor attention and social network following. Random forest model performs best in prediction returns, out of sample precision is around 80%.

Finally, trading strategies based on the bubble and crash prediction and technical indicators show Bollinger band can produce better performance.

立刻免費註冊會員觀看加密貨幣網絡視覺化: 馬上註冊去 !

去哪裡觀看加密貨幣網絡視覺化?: 馬上體驗加密貨幣網絡視覺化




  • 找不到回應