基于A股市场的动量策略和反转策略研究

发布时间：2018-02-27 13:04

本文关键词： 传统动量策略反转策略赢家组合输家组合 K近邻增强　出处：《杭州电子科技大学》2014年硕士论文　论文类型：学位论文

【摘要】：上个世纪八十年代，有效市场假设理论不能解释证券市场上出现反应不足和过度反应的异常现象。行为金融学者认为证券市场上出现反应不足和过度反应是因为投资者对于证券市场上的信息的认知存在偏差，证券投资者的反应不足和过度反应造成市场上的证券价格的动量效应和反转效应。无论是发达的欧美证券市场，还是发展中的我国证券市场，均存在动量效应和反转效应。本文以我国A股市场存在动量效应和反转效应为前提，考察传统动量策略和反转策略收益的同时，对传统的方法做出改进探索。考察传统动量策略和反转策略的收益情况时，利用A股市场2008年1月1日至2013年6月1日的日交易数据，对这个时间区间内的入选编制沪深300指数的股票作为研究对象。不考虑卖空机制的情况下，，运用传统动量策略和反转策略，形成期采用重叠取样方法，持有期采用非重叠取样方法，根据股票在形成期的累积收益率的大小构造赢家组合和输家组合，考察投资策略组合的收益情况。为提高投资回报率，本文对传统的动量策略和反转策略进行改进探索。传统动量策略和反转策略在计算形成期股票的累积收益率时，严重依赖于形成期期初和期末的价格，而股票单个交易日的价格更多表现为随机游走，这种度量方法不一定能得到的真正的赢家或输家组合。为消除股票单个交易日的价格随机游走带来的影响，本文尝试性将统计学知识和数据挖掘算法中的K近邻算法与传统动量策略和反转策略结合，形成了简单增强方法和KNN增强方法。实证得出：运用传统动量策略和反转策略进行投资时，动量策略和反转策略均存在正的超额收益；传统的动量策略和反转策略，持有期相同的情况下，月度收益率随形成期的增加呈增加趋势；传统动量策略的超额收益的均值小于反转策略；在改进的方法下，无论是动量策略还是反转策略，两种增强方法均能提高投资回报率，KNN增强方法的效果好于传统动量方法和简单增强方法。
[Abstract]:-20s, The theory of efficient market hypothesis can not explain the abnormal phenomenon of underreaction and overreaction in the securities market. Behavioral finance scholars believe that the underreaction and overreaction in the securities market is due to investors' reaction to the securities market. There is a bias in the cognition of the information on the. The underreaction and overreaction of securities investors cause the momentum effect and reverse effect of the stock price in the market. Whether it is the developed securities market in Europe or America, or the developing securities market in our country, Based on the premise of momentum effect and reversal effect in China's A-share market, this paper investigates the traditional momentum strategy and inversion strategy, and explores the improvement of the traditional methods at the same time. Using the daily trading data from January 1st 2008 to June 1st 2013 in the A-share market, we investigate the returns of the traditional momentum strategy and reverse strategy. For the stocks selected in this time interval to compile the CSI 300 index as the research object, without considering the short selling mechanism, the traditional momentum strategy and the reverse strategy are used, and the overlapping sampling method is used in the formation period. The non-overlapping sampling method is used in the holding period. According to the size of the cumulative return rate of the stock in the forming period, the winner portfolio and the loser portfolio are constructed to investigate the return of the investment strategy portfolio. In order to improve the rate of return on investment, this paper attempts to improve the traditional momentum strategy and reverse strategy, which depend heavily on the price at the beginning and end of the forming period when calculating the cumulative return rate of the stock in the forming period. The price of a stock on a single trading day is more likely to be a random walk, a measure that does not necessarily result in a real winner or loser combination, in order to eliminate the impact of the random walk of the stock price on a single trading day. This paper attempts to combine the K-nearest neighbor algorithm of statistical knowledge and data mining algorithm with the traditional momentum strategy and inversion strategy to form a simple enhancement method and a KNN enhancement method. The empirical results show that the momentum strategy and the reversal strategy have positive excess returns when using the traditional momentum strategy and the reverse strategy, and the traditional momentum strategy and the reversal strategy have the same holding period. The monthly rate of return increases with the increase of the formation period; the average value of the excess return of the traditional momentum strategy is smaller than that of the reverse strategy; under the improved method, the momentum strategy or the inversion strategy, Both methods can improve the rate of return on investment and the effect of KNN enhancement method is better than that of traditional momentum method and simple enhancement method.
【学位授予单位】：杭州电子科技大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：F224;F832.51

【参考文献】