Mean-variance Based on Filtered Single Linkage Cluster
Huang Fei-xue · Forecasting · 2011
Due to the finite length of the asset return time series,the estimation of the Markowitz 's mean-variance portfolio model unavoidably associated with a statistical uncertainty.So we use the single linkage cluster algorithm to filter the correlation coefficient matrix to avoid this.Then we choose the daily data from 2004 to 2007 of the 50 stocks composing the SSE 50 index to compare the new and the original solution of portfolio optimization and find that:(1)the average reliability of the portfolio composed by the new method is 0.067 higher than that of the portfolio composed by the original method;(2)the realized risk and the predicted risk of the portfolio composed by the new method are respectively 0.1 and 3.3 lower than that of the portfolio composed by the original method;(3)the effective size of the portfolio composed by the new method is 0.0321 smaller than that of the portfolio composed by the original method,which proves the new one can better disperse the risk.So we draw a conclusion that the application of the single linkage cluster algorithm can improve the mean-variance Model.