Estimasi Parameter Model Volatilitas Stokastik dengan Metode Bayesian Rantai Markov Monte Carlo untuk Memprediksi Return Saham

Abstract

Parameters of a distribution are usually unknown values, to find out an estimate is made of these parameters. There are two kinds of parameter estimation methods, namely classical method and Bayesian method. Bayesian method was a method that combines sample distribution with prior distribution. To get a random sample is to use a simulation. One of the simulation techniques used in Bayesian method is Markov Chain Monte Carlo (MCMC) method, which is a simulation method for generating random variables based on the Markov chain. This study discusses Bayesian method with MCMC using Gibbs Sampling algorithm. MCMC method with Gibbs Sampling algorithm works to build a Markov chain by recursively sampling from full conditional posterior distribution of each parameter. In this study, Bayesian method with MCMC using Gibbs Sampling algorithm was applied to estimate parameters of the Stochastic Volatility to converge. Stochastic volatility is a concept that allows for the fact that asset price volatility varies over time and is not constant. This model applied for predicting stock returns of PT. Indofood CBP Sukses Makmur Tbk. (ICBP.JK). Based on Stochastic Volatility model obtained, prediction results for stock returns are almost close to the actual data. Benefit of this research is that the predicted value obtained can be used as a reference for investors to create an optimal portfolio.