Stock Portfolio Construction And Research Based On Entropy TOPSIS Method

Quantitative investment theory is an important branch of modern investment,based on Markowitz's portfolio theory,it has experienced many tests of worldwide market fluctuations,quantitative investment,as a quantitative and computerized scientific trading method that can obtain stable investment returns,has gradu-ally attracted more and more investors' attention.More than 50 countries in the world have established stock markets,and the wide application of quantitative technology accelerates the pace of internationalization of stock trading.However,we must face up to the objective risk problems in the stock market,rational in-vestors tend to consider their own circumstances to invest in stocks,they start from massive data,dig out historical rules,make decisions based on mathemat-ical models,and disperse non-systematic risks by building stock portfolios to maximize returns.In this paper,the stock selection model of scoring method is improved based on the orthogonal-projection entropy TOPSIS method,and the portfolio constructed surpasses the performance of the market,The application of this method in the multi-factor model is an innovative attempt in the field of quantitative invest-ment.This paper takes CSI 300 constituent stocks as the research object,chooses January 1,2009 to December 31,2018 as the factor detection interval of the multi-factor model,and January 1,2019 to December 31,2020 as the model backtest interval,through analysis and processing,11 effective model factors,including size,valuation,rowth,finance and leverage,are finally determined.First of all,based on the scoring method,the stocks with the top 30 comprehensive scores are selected to allocate he stock portfolio with equal rights each month.The backtest result shows that the cumulative return of the strategy can outperform the market index,but the curve fluctuates around the return of the CSI 300 in the early stage.The multi-factor model is not stable,the calculation is complex,and the empirical study takes a long time.In order to solve the above problems and improve the accuracy of the stock selection model,this paper changes the scoring mechanism of the scoing method model by improved entropy TOPSIS method,according to the degree of varia-tion in the data,the effective factors were given objective weights,then,combined with orthogonal-projection TOPSIS method,the target proximity of CSI 300 con-stituent stocks relative to the ideal solution was comprehensively evaluated,and the numerator of the projected distance was taken as the comprehensive score of the stock samples,so as to construct the improved entropy-TOPSIS stock se-lection model.The backtest results show that the portfolio grading method is obviously better than the unimproved entropy-TOPSIS model and the scoring model,it can obtain excess returns stably,the Sharpe ratio and Sotino ratio are higher,and the annualized downward volatility and maximum retracement can be controlled within a smaller range,compared with the traditional scoring method,the improved multi-factor model is more robust and has better ability of predict-ing stock returns;In addition,the backtest time of the improved stock selection model is only one tenth of that of the scoring model,so we can effectively sim-plify the stock selection model while improving the efficiency and accuracy.As an objective method,entropy weight method also avoids the influence of subjec-tive factors when weighting factors.In this paper,the entropy TOPSIS method combined with orthogonal-projection is introduced into the multi-factor stock se-lection model,which can provide new methods and ideas for stock investors who want to beat the market.