The Research On Dynamic Optimization Of High-dimensional Portfolio Based On R-Vine Copula Model

With the first appeared Alibaba's Yu'ebao, the year 2013 opened the first year of the national finance, and it aroused universal concept of finance and investment. While entered 2015, the monetary fund income declined because of the plenty of money supply. On the contrary, A-share market has ushered in the spring, the bull market restarted again after all these years. Huge amounts of inexperienced investors poured into the stock market, which leads the Brokerages being crowded with people and the 24-hour endless business of opening securities account by cellphone or internet. With the rising of the stock market, daily trading volume, weekly new investors repeatedly hit the record. But not for long, mid to late June 2015, A-share market slump, many investors suffered great losses.As much as possible so investors should take what kind of strategy to avoid this kind of thing? There is no doubt that portfolio optimization is a magic weapon. This article mainly studied dynamic optimization of high-dimensional portfolio based on R-Vine Copula TGARCH Model and the monte-carlo simulation technique, which can help some individual investors and institutional investor for portfolio management. In addition, it can be used by market intermediary services consultancy to put forward effective suggestion for some investors.This paper proposed dynamic portfolio optimization algorithm by studying predecessors' research results, combined with related theories. It firstly uses ARMA-TGARCH model to fitting the marginal distribution of a single yield, and then uses R-Vine copulas to depict the joint distribution and nonlinear correlation between multiple assets structure, uses monte carlo simulation technology to forecast the portfolio risk and earnings. Finally, combined with the mean-ES algorithm it optimized investment portfolio.To test the reliability and effectiveness of the dynamic portfolio optimization, this paper selected the ten big industry index of CSI All Share Index as investment targets for empirical research. The result shows that the proposed algorithm in this article can measure portfolio risk more accurately and robustly. Portfolio optimization results can be achieved with different risk preference degree of portfolio to achieve different levels of income. In the bull market phase it successfully achieves the goal of the high-risk high-yield, and part of the portfolio can succeed beyond the market returns, while in a bear market it is at a low risk and low loss, high risk appetite, of course, the portfolio will disaster larger losses but still less than the loss of the market. At the same time, this article compared the results with the results of other algorithms, which proves the superiority of the algorithm in this paper.The characteristic of this study is that the number of portfolio selection is larger, which more accord with the actual situation of investors. Besides, it can optimize level positions to control the risk more effectively.