Research On Moving Average Strategy Based On Time-varying Alpha Revenue And Multiple Hypothesis Testing

Although the total market value of China's stock market is increasing,reaching 56 trillion yuan in 2017,since the global financial crisis in 2008,China's stock market has been generally in bad condition,although in 2015 it experienced a big bull market,but from 2010 to 2017,the average annual return of the Shanghai Composite Index is only 2.12%,which is not as good as the interest on fixed bank deposits.The performance of the Beta strategy which is popular previously is getting worse.More and more investors are focusing on the Alpha revenue,especially after the launch of the 2010 Shanghai and Shenzhen 300 stock index futures contracts.Empirical research has gradually increased,and most studies have shown that Alpha revenue exist in China's stock market.The predecessors' research mainly focused on how to select stocks with Alpha revenue.The selection schemes are mainly divided into two categories,one is the stock selection based on the static Alpha value derived from the variant of the CAPM model,and the other is based on historical revenue or relevant financial indicators,and the selected stock portfolio is usually combined with stock index futures to construct a hedge arbitrage strategy to obtain pure Alpha income.This article also uses the Alpha value for stock selection,but the Alpha value used is dynamic Alpha value that everyone pays less attention to.When selecting stocks,it is not simply to judge the size of Alpha,but to select it in combination with the moving average strategy.This paper,combined with the moving average strategy,believes that the significant increase in Alpha means that the value of individual stock begins to stand out and it is worth buying.According to the above ideas,the moving average strategy based on time-varying Alpha revenue and multiple hypothesis testing is proposed.The strategy is as follows: Firstly,use B-Spline to fit the polynomial function of time and estimate the variant of CAPM model to obtain time-varying Alpha revenue(using G samples in estimation).Secondly,use hypothesis test to determine whether the Alpha value of the G period is significantly larger than the mean of Alpha of the past K period,that is,construct a moving average strategy.Thirdly,use the multiple hypothesis test to select all significant stocks simultaneously as portfolios,holding H days and selling for excess returns.In the empirical part,this paper uses the historical stock return data of Shanghai Stock Exchange A-shares from 2010 to 2017 to implement a back testing on the proposed strategy and finds that it can get an average annual income of about 20%.In order to control the loss,two optimization strategies are proposed.In according to the performances of the three strategies,this paper selects the best four parameters combination,and adjusts the usage scenarios of strategy 2 and strategy 3.It is concluded that best strategy usage plan is: clearly rising markets use strategy 3,in other cases,use strategy 2,and all four parameter combinations can be used.According to this plan,the average annual earnings of the strategy from 2010 to 2017 under the four parameters combination are higher than 24%,the highest is 31.84%,which is much higher than the Shanghai Composite Index of 2.12%;and the minimum revenue outside the training sample from January to November 2018 is-8.60%,which is much higher than the yield of the Shanghai Composite Index of-21.74%,indicating that the strategy of this paper is indeed effective.Finally,this paper combines the indicators in Fama and French's five-factor model to split A shares in the Shanghai stock market and calculate the revenue of each group.It is found that the excess return of small-cap stocks is higher than that of large-cap stocks,and the excess return of stocks with high book-market-value ratio is higher than that of stocks with low book-market-value ratio.The excess return of high-profit stocks is higher than that of low-profit stocks,and the excess return of conservative stocks is higher than that of radical stocks.The conclusion is basically consistent with the traditional view.But this paper finds that under the two-factor grouping,except for the medium book-market-value ratio in the large market value stocks,other medium-level groups get the highest revenue.