Research On FOF's Portfolio Optimization And Risk Management Based On Social Network Methodology

FOF derived from the United States in 1990s,developed rapidly after 2002.The scale of FOF has accounted for about 10%of the total scale of mutual funds so far.Meanwhile,FOF has experienced a bright development in Europe,Asia and some other places.Having witnessed the rapid expansion of Chinese capital market,A-share has become the world's second largest capital market,and domestic mutual funds grow fast as well..At present,private funds management firms and asset management subsidiaries of securities companies are two main issuers of FOF.It is predicted that mutual funds will become another main issuer of FOF since the restriction to FOF has been removed by the Fund Law.FOF provides a valid mersure of asset allocation,and the fund manager can construct the fund portofolio based on accepted risks and specific investment objectives.Along with the rapid development of FOF,the approach of FOF portofolio construction,investment objectives implementation and effective risks management have been an important research theme in theory and practice.Based on social network and modern portfolio theory,efficient market hypothesis and behavioral finance theory,the paper make a full research on FOF from three perspectives of fund evaluation and style classification,FOF portfolio construction and fund risk contagion and management The author combines the theoretical model and the empirical study in order to provide a quantitative model for the issues mentioned above in China.In terms of the study in fund evaluation,this paper establishes fourteen evaluation indicators based on four dimensions of income indicators,statistics indicators,risk indicators and comprehensive indicators.Six principal component indicators which contain 91.7 percent of efficient information of the original indicators are confirmed by principal component analysis(PCA)in which the single indicator and principal component scores based on eigenvalue weighted are respectively used to evaluate funds.Results show that,within a single evaluation indicator,the fund whose value is close to the index ranks higher,while the fund whose value is far from the index ranks lower;the comprehensive principal component scores based on eigenvalue weighted take the whole evaluation indicator system into consideration and fully presents its function.Among these funds,bond funds rank the highest.Besides,in the study of style of funds,the paper makes a clustering analysis on the network of funds with adaptive affinity propagation clustering(APP clustering)using six principal components as measure.We go into part of classification result and find that,in most cases,each cluster contains different styles of funds and they are distributed in different clusters.The result indicates that actual styles of several funds are inconsistent with the counterparties provided in funds' prospectuses though the styles of open-end funds appear resemble in China.The investor shall pay particular attention to this situation when planning to invest into funds.In terms of the study of FOF construction,this paper establishes the network of price for all funds based on social network theory.We make a clustering analysis on the network of funds with adaptive affinity propagation clustering(APP clustering)and screen out the object of FOF from effective funds with two means of center of clustering and index of information gain entropy to construct portfolio optimizations for FOF by mean-variance model based on social network block.Results show that the fund market is an efficient social network where funds networks stretch as a shape of tree,which indicates that some parts of peripheral funds have strong ties to the center funds.In other words,FOF has strong partial clustering features.We screen out FOF with two means of center of clustering and entropy index of information gain and confirm the weight for investment by mean-variance model based on social network block.Results show that the rate of return of 47.6 percent comes from 35 funds selected by information gain entropy strategy rises 3.52 percent by the rate of 46 percent comes from the portfolio by center of clustering.Besides,according to the rolling check of advanced index,results are robust.In terms of the research on fund risk contagion and management,this paper construct a fund network selecting the major stocks as connections to research the existence of microstructure of FOF network and the effect on momentum effect of the major stocks stocks.Results show that fund networks do exist and the connection between the change of fund position and risk contagion is negative;the fund managers who hold the major stocks stocks for the first time have great influence on fund network.After studying funds grouped by residual network density,we find that the momentum effect of stocks with high residual density exist as well;the momentum effect of stocks with low residual density is relatively weak.Besides,this paper announce the mechanism of FOF risk contagion,which expounds that transaction behavior increase the inversion of stocks' return.In addition,the density of fund networks accelerate the process of information transfer for stocks.The paper discusses and surveies deeply the portfolio optimization and risk management for FOF in theory and in practice,which is meaningful in reality.It's shown that FOF has become an important investment instrument and is favoured by investors.Also,FOF is able to provide steady fund flow and further improve investors'structure of capital market,which is also meaningful to Chinese capital market.Based on these results,we suggests the regulator encourage funds management firms to carry out FOF business.It also suggests increasing the types of funds for the purposed of investment diversification and FOF portfolio management.Fund management firms shall strengthen their abilities of investment and research in quantitative field,in order to promote the rapid and scientific development of FOF.However,due to the limitation on the capacity of knowledge reserve and data collection,some further researchs will be studied in the future.For example,how to Optimize FOF's portfolio under different circumstance;how to construct fund evaluation indicator systems with differet industries;how to optimize the weights of FOF's portfolio;how to improve the fund risk contagion model and so on.All these mentioned above will be improved in following research.