Stock Portfolio Decision Based on Cluster Analysis and Principal Component Analysis

Abstract

When determining the stock portfolio, it is necessary to consider factors such as the company's financial indicators, stock price trends, and macroeconomic conditions. In order to simplify the process, the clustering method is used to classify the company according to the financial indicator data, so as to select stocks when diversifying the investment, and then Sharp Ratio is calculated to conduct risk assessment on the selected targets. Finally, the principal component analysis method is used to simplify the obtained data, and then predict the future trend of the stock price by linear regression, and finally an ideal investment portfolio is determined. Introduction Among many financial products, stocks are the targets of many investors. The stock market is a very important part of the investment market. As we all know, options, bonds and funds are financial instruments based on stocks. Therefore, the fluctuation of stock prices and the overall situation of the stock market have been widely focused. Before investing in the stock market, investors will observe the selected targets for a period of time with some methods. These methods are generally K-line observation or fundamental analysis. But the stock market is affected by many factors, and more accurate methods are needed to determine their investment targets in order to obtain more stable returns. One of the important premises of portfolio theory is “diversified investment”, which means we should diversify the capital into different types of companies to avoid risks and maximize the return. But sometimes it’s hard to define “different type”, so here we can select the financial indicators that reflect the company's situation and use clustering to classify them. Deng Xiuqin also used the clustering method in her article [3]. here we use a similar method to classify several selected targets, in Deng Xiuqin's article, only selected five indicators of profitability to reflect the company's type, and in this paper, we have made some adjustments to the financial indicators selected during clustering, taking into account the ability of a company's profit, debt repayment, and growth, therefore obtain more reasonable classification results. After classifying it, analyze the previous data for different categories, and obtain the Sharpe Ratio [4] of each underlying stock to determine the size of the investment risk, which is helpful for us to choose stocks. It can help us choose a higher-yielding and more stable investment target within the acceptable risk range. In the selected targets, the principal component analysis method is used to analyze the indicators reflecting the price changes, and several representative principal components are obtained. According to the results obtained in < Stock Price Forecast Based on Principal Component Analysis and Generalized Regression Neural Network> written by ZhuoXi Yu. [5] Based on Zhuoxi’s paper, this article deletes the three indicators of earnings per share, return on equity, and net assets per share, because the operating conditions in the first part of the cluster analysis has been used the company's financial indicators, this is only part of a number of technical indicators to predict future stock price, and ultimately determine an appropriate investment portfolio based on the results. 6th International Conference on Management Science and Management Innovation (MSMI 2019) Copyright © 2019, the Authors. Published by Atlantis Press. This is an open access article under the CC BY-NC license (http://creativecommons.org/licenses/by-nc/4.0/). Advances in Economics, Business and Management Research, volume 84