Modeling and Forecasting Jakarta Stock Exchange: Stock Market Volatility

Abstract

Various techniques have been introduced by earlier academicians, from random walk model to GARCH model with its univariate and multi-variate derivations. The models have proven their strength and weakness in forecast accuracy and mean square error (MSE). On this paper, we try to conduct simulations on forecasting model for capital market. The research is aimed at finding forecast model that fits Jakarta Stock Exchange (JSX) the best. JSX is used as sample, which represents real circumstance of emerging market (microstructure).In general, outlier data is found on JSX return as the consequence of many events causing data shock. We try to compare mean model with volatility model to examine capability of a model in predicting return. Finally, we find that combination of autoregressive (AR) model, and the addition of regime independent variable, which is represented by a dummy variable, is suitable to JSX data return characteristic. This, in turn, can increase the value of R2-adj and be therefore meaningful. At the end, the assessment on the relationship between return and risk reveals that there is positive relationship between market return and the associated volatility.