The Optimal Portfolio Decision Based On Thedeviation Of Excessive Loss

The global finance systems integrate gradually along with the development of internet and information technology. The financial interaction among countries are getting more and more frequent, which brings new opportunities on global economy as well as new challenges impacted by financial crisis just as the "butterfly effect". The Asian financial crisis in1997 and US subprime crisis in 2007 threw us a profound warning that the global financial markets are getting more and more turbulent under economic globalization. The financial crisis in one country or district would quickly diffuse its impacts into the whole world. In financial practices, the small-probability extreme events emerged frequently and impacted excessive loss upon world economy and global financial systems, which makes it substantially critical for financial risk prevention. In this paper, we will investigate the optimal portfolio decision by minimizing the deviation of excessive loss with small probability.We firstly introduce the significance of this research and put forward an outline of the contents, ideas, and innovations of this research based on a detailed survey on the development of portfolio theory.In Chapter 2, we investigate the evolution of portfolio selection models and put forward a portfolio model to minimize the deviation of excessive loss with small probability, which takes into account of the investor’s psychological characteristics of risk-averse and the small probability event of excessive loss emerged frequently in financial practices.In Chapter 3, we firstly choose 131 stocks from 28 industries in Shanghai and Shenzhen Stock Exchange by fundamental analysis of 12 indices, and 11 stocks are then chosen to construct a core asset set by clustering analysis technology. Next we propose a scenario generation approach based on principle component analysis, which considers the asymmetric distribution of asset returns and the excessive loss events. This approach succeeds in escaping from the "dimension curse" by generating less and robust scenarios.In Chapter 4, we test the validity of scenario generation approach and the portfolio model proposed in this paper by constructing the optimal portfolio decision with 11 stocks selected by cluster analysis. The numerical simulation shows that both are valid and effective. This portfolio model exhibits its advantages than Conditional Value-at-Risk( CVa R) model in forecasting tests and achieves its aim to control the deviation of excessive loss.Finally we conclude this paper and look into future research in Chapter 5.