Stock Trend Prediction And Stock Selection Strategies Research Based On DTW-KNN With Volume-Price Relationship

With the development of artificial intelligence technology,quantitative invest-ment has become the focus of the development of European and American capital markets.More and more companies apply data mining technology and machine learning algorithms to financial quantitative investment areas.The use of ma-chine learning methods for stock price and trend forecasting,as well as the use of stock future trends for quantitative stock-picking research is very practical and has a good development prospects.This paper bases on the K-Nearest Neigh-bors algorithm(KNN)which is a machine learning method to predict the trend of stock price and stock index.The dynamic time warping(DTW)applies to measure time series'similarity as the sub-process of the classification algorithm.This paper designs DTW-KNN model and fully analyzes the relationship between volume and price,use the turnover rate as the measure of volume,and constructs an improved model based on the volume-price relationship.At the same time,the Support Vector Machine model(SVM)is used as the contrast model.In this paper,the three models are used to the Sino-US stock market to respec-tively forecast trends in the future of the SSE 50 Index and the S&P 500 Index.The empirical results show that the SVM model and the DTW-KNN model and the improved model based on the volume-price relationship can achieve more than 55%accuracy of stock index trend prediction,indicating that all three models are effective.And the performance of the improved model is significantly better than the DTW-KNN model and the SVM model.Based on the prediction of stock trend and the actual situation of Chinese stock market,this paper builds two quantitative stock selection strategies.The results of strategy backtest show that the performance of the stock selection strategies based on DTW-KNN model and improved model based on volume-price relationship are better than the performance of SSE 50 index.The annualized rate of return of the strategies respectively reached 16.8%and 19.7%and both strategies are able to obtain excess returns.