Timing Strategy Of HS300 Index Based On Wavelet Denoising And Random Forest Algorithms

With the wide application of big data methods and the rapid development of quantitative investment,more and more experts and scholars have tried to combine big data methods with stock market forecasts,and index trend prediction is undoubtedly the most meaningful analysis topic.Not only can it be used for market research,but also convenient arbitrage methods.From the practical point of view,a large number of studies show that the integrated class algorithm performs significantly better than the single classifier in the prediction of Jinrongshichang.But at present,most index trend prediction studies based on integrated algorithms tend to focus only on technical indicators,and ignore the linkage between national stock markets and the impact of various types of assets.In addition to the commonly used technical indicators,this paper also includes the main international index,the base difference between futures and spot,interest rate market,exchange rate and other indicators into the characteristics of the Shanghai and Shenzhen 300 index,and establishes a random forest model on this basis,and then carries out index prediction.On the other hand,it is generally believed that the stock price trend consists of long-term trends and short-term disturbances.The uncertainty of short-term disturbances undoubtedly makes it difficult to predict stock price trends.In some studies,the average line is used to describe long-term trends,but the problem is that the average line has a certain lag.This paper uses the wavelet de-noising method commonly used in signal processing to filter out short-term disturbances,so that the long-term trend can be reflected.The index trend after wavelet de-noising is one of the characteristics of random forests.This paper first introduces the theoretical basis of wavelet de-noising and random forest algorithm,then selects 5 categories,a total of 30 indexes to train the model.And within 300 trading days modeling,back test,and according to the results of the back testto build a multi-empty combination.The results show that the changes of technical indexes,other important international stock indexes,futures and exchange rate have indeed played a role in predicting the rise and fall of the Shanghai-Shenzhen 300 index.The trend after wavelet de-noising is also one of the important characteristics of predicting the rise and fall of the Shanghai-Shenzhen 300 index.The prediction accuracy of the model was 57 % in 300 trading days.According to the model,the multi-empty combination yield is significantly higher than the index.