A controllable portfolio stock selection method based on an AP clustering algorithm

Abstract

The invention discloses a controllable portfolio stock selection method based on an AP clustering algorithm, The steps include: acquiring stock pool data for a fixed period of time, calculating a return rate of each stock in the acquired stock pool over the fixed period of time, setting controllable parameters, wherein the AP clustering algorithm is used to cluster the time series, and the similarity matrix, attraction matrix and belonging matrix of each stock in the stock pool are calculated, and the clustering results are obtained, and judged by combining the controllable parameters until the clustering results satisfy the controllable parameters. According to the clustering results, the clustering centers in each cluster are selected as the investment portfolio for investment analysis.By adopting controllable AP clustering for stock selection, the invention greatly controls the problem that the number of clusters in the AP clustering is too much / too little to carry out correct portfolio, improves the independence of each stock in the selected portfolio and reduces the investment risk.

Description

Technical field [0001] The present invention relates to the field of investment strategy, and more specifically, to a controllable portfolio stock selection method based on AP clustering algorithm. Background technique [0002] Investors invest their assets in two or more financial products as investment portfolios. The investment portfolio idea is to diversify the wealth of investors into multiple shares to reduce the investment risk of investors, so that investors will not cause unnecessary losses due to extreme market fluctuations. In the selection process of the investment portfolio, due to the large number of individual stocks and the greater impact of the stock market correlation, it is not easy to find the optimal portfolio candidate stocks in the process of stock selection. Therefore, it is impossible to circumvent the factor one in the actual investment portfolio. The loss of assets caused by the risk of multiple shares affecting multiple shares is a huge challenge for ...

AI Technical Summary

Problems solved by technology

The research in recent years is basically based on the traditional learning method of the traditional investment portfolio, through stock selection research on investor preferences and market fluctuations, but does not consider whether the selected stock portfolio is in line with the real investment value, And whether it has the ability to reduce risks; and the research process is based on traditional statistical methods, and does not consider the methods brought by machine learning in reality, so portfolio research is still in a preliminary stage and cannot be really applied to the market

Images