Research On Stock Index Trading Strategy Based On FASVR-FWKNN-MACD Combination Model

In quantitative investment,investors hope to use a certain tool or strategy to predict the trend of stock index,so as to obtain stable investment income.However,due to many factors that cause the stock index to change,it is difficult to predict.How to better predict the trend of stock index points is a difficult and important issue for investors.Based on historical stock index data,using mathematical models to predict trends can overcome the subjectivity of human judgment and capture more reasonable trading signals.In order to solve the above problems,this paper constructs the combined model FASVR-FWKNN-MACD to predict the change trend of the stock index closing point,and designs the trading strategy based on the model,and then uses the designed trading strategy to conduct empirical trading on the five stock indexes.This paper is divided into four parts from the research process:First,the paper first uses the stepwise regression method to determine the input variables of the support vector regression machine(SVR):the number of opening points,the highest points,the lowest points,the closing points,the trading volume,and the transaction amount.Second,build the FASVR-FWKNN model,and use the model to predict the closing number of the first day of the future based on the previous day's stock index data.In the meantime,the parameters of SVR are optimized based on the firefly algorithm,and the parameter enhanced support vector regression machine(FASVR)is obtained.Then the weighted K-nearest neighbor algorithm(FWKNN)is used to correct the prediction error of FASVR.Third,due to the low accuracy of FASVR-FWKNN's ups and downs prediction,this paper uses the medium and long-term prediction of the number of closing points of the same-same moving average(MACD)stock index to filter the short-term forecast of the FASVR-FWKNN model,and obtain a stronger forecast signal..Thus,the combined model FASVR-FWKNN-MACD is constructed and the stock index trading strategy is designed based on the combined model.Fourth,use the designed trading strategy to conduct empirical trading on five stock indices.The results show that the trading strategy based on FASVR-FWKNN-MACD has higher yield and success rate than other trading strategies,among which the yield is between 6.2%and 21%,and the winning rate is between 56.2%and72%.The main innovations of this paper include:First,the paper also uses the FASVR-FWKNN regression model to predict the ups and downs and the forecast of the rise and fall.The model's ups and downs direction prediction and the technical indicator MACD's ups and downs forecast combined with the trading timing;different from the usual fixed percentage stop loss and fixed point stop loss,this paper based on FASVR-FWKNN forecast for the first day of the next day's closing point And the test set prediction error to set the mobile stop loss point to further control the trading risk.Second,this paper uses the medium and long-term forecast to filter the short-term forecast method to predict the rise and fall of the stock index closing point,and also the combination of technical indicators and mathematical models.This paper firstly uses the combined model FASVR-FWKNN to forecast the ups and downs of the stock index on the first day of the future,and then uses the medium and long-term trend forecast of the technical indicator MACD to filter the short-term forecast of FASVR-FWKNN to select a stronger signal.