Research On Quantitative Timing Strategy On Flow Of Capital

With the progress of intelligent algorithm and data mining technology,people are more and more scientific in dealing with big data.Meanwhile,quantitative investment based on big data has begun to shine in the field of securities investment.Avoided the hot area of quantitative investment,this paper takes the market low frequency data as the research object,and constructs a timing strategy based on the capital flow for the medium participants in the retail market.The core strategy is divided into two parts,stock selection and market timing;the stock part,different from the traditional quantitative stock selection strategy in order to obtain high income,the paper's model take the high volatility,and abundant information stocks for the purpose;secondly,the timing part of this models based on the flow of funds index,select large caliber as extracting data,to obtain the target with potential good returns,select optimal speed line combination,at the same time,using the classical theory of MACD,KDJ,judge the trading point.Stock selection process;this article takes the Shanghai and Shenzhen 300 index constituent stocks as the original research object through the Wind consulting platform,after preprocessing the original data,282 valid stocks are obtained.Then,the correlation matric C is calculated by using the corresponding logarithmic return time sequence data of the complex network construction window period(2016/01/01——2016/12/31),and through distance transfer equation make it into distance matrix Cd;after that,make effective stock is the node of the complex network,and the distance between the stocks is the side length,by setting different link thresholds,obtained complex networks with different structural characteristics,and analyzed the statistical properties of complex networks under different link threshold,and found that the constructed network had good robustness.Finally,the complex network corresponding to the link threshold at interval 0.276—0.336(0.001 step strength)was selected as the complex network.The research object,a total of 60 complex networks,respectively selected each complex network,the degree centrality the centrality of centrality,the top five of the top rank of the centrality,and the stock pool at the time ofchoice,the 60 complex networks eliminate the repeated stocks 12 of superior stocks,the tight centrality excellent stock 12,and the final total get 34 stocks when the stock is selected.Time traction processing:this paper selects the rolling window span of the fast and slow line between the 1-29 days and gets 406 fast and slow lines.Then,the theoretical odds,the daily transactional probability,the relative yield,the loss rate index of the transaction results are limited by the model training window(2017/01/01——2017/07/31)and the different fast and slow line combination.Finally,after a quantitative selection based on constructing the model in the training window(2017/01/01——2017/07/31)and model of measuring window period(2017/08/01——2018/03/01)of the simulated trading firm,models are better proved by annual rate of return,the sharp ratio,the volatility ratio and maximum retracement ratio,than those of the Shanghai and Shenzhen Composite Index(without transaction costs)benchmark investment income obtained.Especially in the down ward trend of the market are more prominent in the model strategy and have a better ability to avoid the downside risk of the market.At the same time,the model has no transaction cost constraints at almost every time period.The performance is always better than the constraint of the transaction cost,and as time goes on,the transaction cost will accumulate on the model.The impact is constantly magnified,especially on the two indicators of the information ratio and alpha value of strategic yield significantly.Therefore,it is proved that the low frequency data of tour stock market in time window period(2016/01/01——2018/03/01)is effective to construct the quantitative strategy,and it also proves that the stock market in our country is invalid at the present stage.Proved that the effectiveness of low frequency data in the construction of quantitative strategy in China's stock market,and it is also proved that China's stock market is ineffective.