管理评论

管理评论 ›› 2024, Vol. 36 ›› Issue (6): 94-106.

• 经济与金融管理 • 上一篇    

基于处置效应及动量效应的加密货币市场投资策略

刘帅1,2, 房勇1,2, 汪寿阳1,2   

  1. 1. 中国科学院数学与系统科学研究院, 北京 100190;
    2. 中国科学院大学经济与管理学院, 北京 100190
  • 收稿日期:2021-02-08 发布日期:2024-07-05
  • 作者简介:刘帅，中国科学院数学与系统科学研究院、中国科学院大学经济与管理学院博士研究生；房勇(通讯作者)，中国科学院数学与系统科学研究院、中国科学院大学经济与管理学院副研究员，博士生导师，博士；汪寿阳，中国科学院数学与系统科学研究院、中国科学院大学经济与管理学院研究员，博士生导师，博士。
  • 基金资助:
    国家自然科学基金项目（71631008）。

Investment Strategy of Cryptocurrency Market Based on Disposition Effect and Momentum Effect

Liu Shuai1,2, Fang Yong1,2, Wang Shouyang1,2   

  1. 1. Academy of Mathematics and Systems Science, Chinese Academy of Sciences, Beijing 100190;
    2. School of Economics and Management, University of Chinese Academy of Sciences, Beijing 100190
  • Received:2021-02-08 Published:2024-07-05

PDF (PC)

55

摘要： 针对加密货币市场，本文选取资本利得量作为处置效应的代理变量，基于公开数据并采用计量方法实证检验了加密货币市场中存在处置效应和动量效应；然后依据处置效应和动量效应，针对仅含加密货币这一类资产而不含其他传统资产的投资池设计了三类投资策略。本文的投资标的不仅含有市值较大的比特币和以太坊等某几种加密货币，还选取了更多种类的货币，这些货币总市值占全部加密货币的95%以上，种类更丰富。在各种参数设置下，结合了两种效应的投资策略均表现优于单独基于某一效应指导设计的策略，并且基于投资者行为效应的三类投资策略均优于等权重模型，具有稳健性和普适性。特别地，本文还对加密货币市场中动量效应和处置效应的持续时长、资本利得量的参数选择和动量策略的排序期参数设置进行了讨论，发现：加密货币市场较股市的变换更快，迭代周期缩短很多，仅存在两周的动量效应，处置效应对收益率的影响不超过一个月；须选用日数据或更高频数据计算资本利得量并进行处置效应存在性检验的回测才有效，选用低频的周数据无效；动量策略的排序期超过动量效应持续时间的一半时，动量策略受排序期参数的影响不大。

关键词: 加密货币, 处置效应, 行为金融, 资本利得量, 投资策略

Abstract: This paper selects capital gains overhang as a proxy variable for disposition effect, using public data and measurement methods to empirically test and confirm the existence of disposition effect and momentum effect in the cryptocurrency market. Three types of investment strategies have been designed for an investment pool that contains only cryptocurrency assets and no other traditional assets under the guidance of disposition effect and momentum effect. The investment targets of this paper are not only some cryptocurrencies with large market capitalization, such as Bitcoin and Ethereum, but more varieties, accounting for more than 95% of the total market value of all cryptocurrencies. Under various parameter settings, the investment strategy that combines these two effects outperforms any strategy based on a single effect, and the three types of investment strategies directed by investor irrational behavior guidance are all better than the equal weight model which are robust and universal. In particular, this paper also discusses the duration of the momentum effect and disposition effect in the cryptocurrency market, the parameter selection of capital gains overhang and the ordering period parameter setting of the momentum strategy. It is found that the cryptocurrency market changes faster than the stock market, and the iteration cycle is shortened a lot. In addition, the momentum effect lasts no more than two weeks and the impact of disposition effect on the rate of return are within half a month in the cryptocurrency market. It is only valid to use daily data or higher frequency data to calculate the amount of capital gains overhang and to conduct back-testing to test the existence of disposition effect. On the contrary, the use of low-frequency weekly data is invalid. When the ordering period of the momentum strategy exceeds half of the duration of the momentum effect, the momentum strategy is not significantly affected by the ordering period parameter.

Key words: cryptocurrency, disposition effect, behavioral finance, capital gains overhang, investment strategy