The Study Of Stock Investment Policy Based On Local Similarity Analysis

With the development of economy and the living standard improves,more and more people start to investment and finance,such as applying extra money in stocks,bonds,funds and other financial products.There is no doubt that the stock market is the hottest market in recent few years.Whether retail investors or institutional investors,are eager to make a fortune in the stock market.We found that a kind of covariance phenomenon often occurs in the stock market,where the correlation between stocks is very high.The study of stock market linkage can not only help investors to analyze stock information and grasp market dynamics,but also help investors to make a reliable investment strategy in order to increase revenue.For now,the research about stock linkage rarely involves between individual stocks,but more focuses on the region or the plate.In order to explore the correlation between the individual stocks,this paper applies the local similarity analysis method to study the similar interval and the delay between stocks.This paper introduces local similarity analysis with time delay,which is applied to the stock market to analyze the similarity of individual stocks and obtain the local similarity interval and time delay of two stocks.Our data set is the closing price of 90 stocks in Shenzhen Stock Exchange from January 4,2013 to December 30,2016.We chose the stock price from January 4,2013 to December 31,2015 as training dataset and the other as test dataset,and investigated local similarity of training dataset.Among the 4005 samples,we selected 787 samples with D? 0.During the 787 samples we choose those with time delay 4 ? |D| ? 20 and similar length greater than 80%data as the investment strategy research samples.The core idea of the investment strategy is to establish the trading model according to the property of the two stocks with time delay.In this paper,we get the optimal parameter from the test dataset,and apply them to test dataset to vetify the effect of our model.The result shown as follows:1.There is a strong correlation between stocks in stock market which increase and decrease simultaneously.2.After obtaining local similarity interval I =[i,i + l-1]and J=[j,j + l-1]and the time delay D by local similarity analysis,we find that there were 490 pairs(62.26%)with similar intervals in more than 80%time points,of which 344 pairs(43.71%)length is more than 90%in Shenzhen Stock Exchange samples with time delay D ?0,and 712 pairs(91.04%)with similar intervals in more than 80%time points,of which 440 pairs(56.27%)length is more than 90%in Shanghai Stock Exchange samples.3.The investment strategy model based on local similarity analysis method achieves a good income in 2016.The average rate of return reach 12.78%in Shenzhen Stock Exchange samples,while the rate of Shenzhen securities Composite Index in 2016 is only-12.46%.The average rate of return reach 4.93%in Shanghai Stock Exchange samples,while the rate of Shanghai securities composite index in 2016 is only-5.84%.Our strategy model in test set always beats the market.