The Optimization Of MACD Timing Strategy Based On Machine Learning Methods

In the field of stock market prediction,predicting the future price or trend is one of the most challenging problem.In recent years,with the development of machine learning technology,in the related research of stock price prediction,more and more people have introduced machine learning models into traditional financial analysis technology and combined machine learning technology with traditional technical analysis indicators to predict the trend of the stock market.This paper proposes a hybrid technology analysis trading strategy based on machine learning,which optimizes the buying or selling trading signals generated by traditional methods through machine learning technology so as to improve income.The innovation of this method is to use the machine learning method to predict the future stock price,and define and calculate a new trend index,which is combined with the technical analysis index for optimization.In order to choose the most suitable machine learning technology,this paper introduces and analyzes the prediction effect of three methods:support vector regression,random forest,and artificial neural network in the stock market.MACD timing strategy is selected as the basic technical strategy of this paper.At the same time,for the selection of MACD parameters,grid search is used to calculate the optimal parameters.Since the outbreak of COVID-19,China’s stock market has been affected by many factors.In order to objectively analyze the effect of the hybrid trading strategy proposed in this paper on the improvement of investment income,this paper divides the empirical research into two stages:the pre-epidemic stage and the post epidemic stage,using the daily trading data of the three major domestic indexes,SSE 50,CSI300 and CSI500 to backtest.Firstly,SSE50 index is taken as an example.This paper compares the prediction accuracy of three machine learning methods and selects the random forest method as the optimal stock price prediction method based on the prediction results,then uses the data of 2018 and 2020 to calculate the optimal MACD parameters respectively,finally backtests and analyzes the profit effect of the optimization strategy,and also backtests the strategies of CSI 300 and CSI 500.The result shows that adding the machine learning trend index to the technical analysis strategy can improve the transaction signal and the competitiveness of the technical analysis strategy.This strategy can seize the opportunity and greatly increase profits when the market is good.When there are many influence factors and large fluctuations in the stock market,this strategy can also effectively avoid the price shock area and reduce the risk.