Font Size: a A A

Comparison Research And Application Of Portfolio Models Based On CVaR,WCVaR And DRO Method

Posted on:2024-02-21Degree:MasterType:Thesis
Country:ChinaCandidate:J J ZhengFull Text:PDF
GTID:2530306923454424Subject:Applied statistics
Abstract/Summary:PDF Full Text Request
Since the yield data in the financial market usually have the characteristics of peak and thick tail,and the traditional financial theory assumes that the return on assets follows the normal distribution,the mean-variance based portfolio model proposed by Markowitz in the 1950s cannot fully measure its distribution characteristics.It is an accepted fact that the return on assets data do not meet the normal distribution.In this paper,the distribution of daily stock return data is initially judged by the histogram,and Jarque-Bera normal distribution test confirms that the data does not conform to the normal distribution.Therefore,this paper selects three portfolio models based on CVaR,WCVaR and DRO constraints to test the applicability of the optimization model in domestic and foreign markets.In order to compare the optimal portfolio strategy given by these three portfolio models in the same period of time,this paper draws the cumulative daily return chart of the portfolio for comparison.The results show that the optimal portfolio strategy given by the portfolio model based on DRO constraints has certain advantages no matter before and after the epidemic or in the stock market at home and abroad.However,the portfolio yield based on CVaR model is too low,and the variance of the portfolio is not the lowest.In the empirical part,this paper considers the portfolio strategy based on CVaR constraint when the probability distribution of random variables is known,and the portfolio model based on WCVaR and DRO constraint with only partial information given by the probability distribution of random variables.To obtain the performance of the optimal portfolio through the historical stock return data,this paper takes the equal-weight portfolio(Eq_model)as the reference benchmark,and compares the effectiveness of the three models in different stock trading markets in the same period of time.The Sharpe ratio can be used to quantitatively evaluate the returns and risks of portfolio strategies,which is intuitive and explanatory.In this paper,the Sharpe ratio of the above three models in different time periods is calculated.The results show that the Sharpe ratio of the portfolio model based on DRO constraints is almost the highest in all scenarios,but the variance of the portfolio is also large.The reason is that the optimal portfolio weight usually gives a very high weight to a stock,and attaches great importance to portfolio yield,but the ability to use portfolio to disperse risk is poor.Through comprehensive comparison from various perspectives,this paper holds that the portfolio model based on CVaR and WCVaR constraints makes full use of the portfolio to diversify the risk,but the model obtains the optimal portfolio cumulative yield lower than the portfolio model based on DRO constraints.
Keywords/Search Tags:Investment Portfolio, Conditional Value at Risk, Worst-Conditional Value at Risk, Distributionally Robust Optimization, Stock market
PDF Full Text Request
Related items