SSE 50 Index Change Forecast Based On SVM

With the rise of China’s great power,economic development is changing with each passing day,More and more people hold the wealth in their hands,A variety of capital preservation financial products on the market after another,Residential property has become the property of investment,All of the above show that China has entered an era of investment and wealth management.While bank deposit rates are relatively low,Ordinary people find it hard to resist the currency devaluation caused by inflation.In previous years,most of the middle class choose to invest in real estate for capital appreciation.However,the central government in order to prevent soaring real estate prices so sharp that financial systemic risk,In recent years,a series of policies have been put in place to curb real estate bubble speculation.So many wait and see think that even now to the next 5 years,the real estate market is not very good investment returns.at the same time,Since its establishment in 1990,the stock A-share market has been growing rapidly for 30 years,In the country is more corporate financing and investment channels for investors.on the other hand,The country firmly suppressed real estate bubble speculation.Therefore,in recent years,because of the crowding-out effect of real estateMore money began to flow to the secondary market.The regulation of China Securities Regulatory Commission is increasingly tighter,prohibition of various themes hype.Begin to gradually guide the market funds for value investment.With the opening of Shanghai-Hong Kong Stock Connect and Shenzhen-Hong Kong Stock Connect,and Ming Sheng announced in June 2017 that A shares will be included in the MSCI Emerging Markets Index.All show that A-share market will be in the near future and the international market mature,and radually eliminate retail investors,The future market can only invest in good fundamentals of blue chips.SSE 50 Index,This index can better represent the A-share market in the future direction of investment,This paper divides the movements of the SSE 50 Index into two categories(up,down),Through the stock index last trading day index to predict the next trading index changes.The first chapter of the article introduces the background and significance of the topic,and research situation of stock market at home and abroad.The second section describes statistical learning theory,The third chapter introduces the stock market commonly used indicators,The fourth chapter is the empirical study of the model.Applying the normalization method to pretreatment the 19 technical indexes of the 362 trading days of the SSE 50 Index,uniform dimension,find the relevance,Then,five principal component variables are obtained by principal component analysis and dimensionality reduction.Then these five principal component variables as input vector,using non-linear support vector machine to train 290 trading days of data,Select Gaussian kernel function,Through the grid parameter optimization and cross-validation method to get the best nuclear parameters,making the classification effect is good,get a function model,Finally,we use the five principal components of 72 test days in the test set to input the model,Get the next trading day ups and downs.Combined with specific indicators of data,The SVM model verifies the classification accuracy of the SSE 50 index.Obtained the result of SVM model earnings outperforming SSE 50 index over the same period.However,due to limited technology and lack of experience,This article failed to get a better classification accuracy,Hope that the future can be combined with integrated portfolio approach,Get a higher classification accuracy,better classification prediction model.on the other hand,The frequency of sample data selected in this paper is based on the daily forecast period,Shorter time,greater volatility,noise is also great,The daily uncertainty may not be included in the forecast system,Eventually resulting in prediction error,So follow-up can be weeks,months of the sample test.