文章使用2013年1月-- 2017年6月的日度比特币交易数据,对序列使用马尔科夫区制转移模型(参数方法)、小波变换(非参数方法),研究比特币收益率在不同尺度分布的定量特征.由MS-AR模型分析结果可知,在低收益率区制下,投资者主要对前一两天以及大概一周的收益率有所关注,在高收益率区制下,投资者的心理在发挥主要作用,普遍认为昨日涨今日跌的事实,因此对比特币进行投资具有极强的投机心理.由小波变换分析结果可以看出,不同时间下,比特币影响因素不同.当时间跨度选取较小时,主要是投资者的投机心理以及``羊群效应"起主要影响,随着时间跨度的增加,达到半年左右时,此时主要是比特币市场的自主调控,因此也会出现小幅度的波动,但当时间跨度达到一年以上,此时政府等监管部门对该市场的管控起主要的作用,使得收益率波动情况达到几乎平稳的状态.对比发现,小波变换能够较好地反映比特币收益率的分布情况.

In this paper, we use the Markov regime switching model (parametric method) and wavelet transform (nonparametric method) to study the quantitative characteristic in different levels of the rate of return of Bitcoin from January 2013 to June 2017. {{From MS-AR model analysis results}}, we can see that under the regime of low rate of return, investors mainly pay attention to the return of one or two day and about a week before. Under the regime of high rate of return, the mentality of investors makes them believed the fact that the return of Bitcoin will fall today if it rose yesterday. So the investors have a strong speculating psychology in the investment of Bitcoin. By {{wavelet analysis}} results, it can be seen that the impact factors of Bitcoin vary at different time. When choosing a short time span, the investor's speculative psychology and the ``herding effect" mainly make difference. With the growth in time span, reaching about six months, it's {{self-regulation}} of the Bitcoin market mainly plays a major role so that there will be a small amount of volatility. But when the time span reachs more than one year, the control of the market from the government and other regulatory authorities plays an important role and makes the rate of return at a steady state. {{The comparison shows that the wavelet transform can better reflect the distribution of bitcoin returns.