MENENTUKAN PORTOFOLIO OPTIMAL MENGGUNAKAN MODEL CONDITIONAL MEAN VARIANCE

Abstract

When the returns of stock prices show the existence of autocorrelation and heteroscedasticity, then conditional mean variance models are suitable method to model the behavior of the stocks. In this thesis, the implementation of the conditional mean variance model to the autocorrelated and heteroscedastic return was discussed. The aim of this thesis was to assess the effect of the autocorrelated and heteroscedastic returns to the optimal solution of a portfolio. The margin of four stocks, Fortune Mate Indonesia Tbk (FMII.JK), Bank Permata Tbk (BNLI.JK), Suryamas Dutamakmur Tbk (SMDM.JK) dan Semen Gresik Indonesia Tbk (SMGR.JK) were estimated by GARCH(1,1) model with standard innovations following the standard normal distribution and the t-distribution.  The estimations were used to construct a portfolio. The portfolio optimal was found when the standard innovation used was t-distribution with the standard deviation of 1.4532 and the mean of 0.8023 consisting of 0.9429 (94%) of FMII stock, 0.0473 (5%) of  BNLI stock, 0% of SMDM stock, 1% of  SMGR stock.