Research On Market Timing Strategy Of Stock Investment

With the gradual deregulation of domestic financial markets and the progress of financial innovation,investors can develop more quantitative investment strategies for stock trading.Market timing is a significant quantitative investment or trading strategy in which market participants attempt to beat the stock market by predicting its movements and buying and selling accordingly.This paper aims to provide a reference for stock traders when making investment decisions,for instance,to capture the correct investment timing and to obtain excess returns.In this paper,we proposed a market timing strategy based on Functional Data Clustering Analysis.First,this thesis adopted the Gaussian Mixture Model-based(GMM)clustering analysis to classify high-frequency historical data and analyzed the representative trends as trading signals.Secondly,this paper also adopted the Partitioning Around Medoids(PAM)algorithm-based clustering method to make predictions on the future stock price movements of each category.Finally,this paper verified and measured the effectiveness and robustness of the strategy.This study found that:(1)This market timing strategy based on functional data clustering analysis can pick up on stock trends successfully and the average win rate exceeded 55%.The empirical analysis showed that these two functional clustering models both optimized the traditional clustering model and improved the market timing strategy;(2)The clustering analysis based on Gaussian Mixture Model has a great effect on the classification of stock price trends,and the results could show the characteristics of "similar attributes within the class and obvious differences between the classes" significantly.The average win rate of long and short position forecasting based on this method is 60.42%;(3)The clustering analysis based on the PAM algorithm is effective in predicting the future movement of stocks.This method showed superiority in the empirical research of stock price trend prediction,whose average win rate is 56.25%.