Non Parametric Kernel in Time Series Data of Composite Stock Pricing: An Application at Indonesia Stock Exchange

Astutik, Yayuk Setyaning (2019) Non Parametric Kernel in Time Series Data of Composite Stock Pricing: An Application at Indonesia Stock Exchange. In: International Conference 2nd Global Conference on Applied Science, Environment & Industrial Engineering (GSEI), 13-14 June 2019, Seoul, South Korea.

Preview

Text 3. Book of Abstract GSEI 2019.pdf Download (1MB) | Preview

Abstract

In this paper, non-parametric kernel time series data is introduced so that it can be used to do IHSG forecasting for several future periods. The Composite Stock Price Index(IHSG) is one of the stock price indices in Indonesia with time series data. We estimate non-parametrical composite stock pricing through the selection of optimum bandwidth by using cross validation with result of 305,1946. Nadaraya Watson’s estimation and Gaussian kernel functions are used in kernel non-parametric regression. We compare multiple regression analysis and kernel non-parametric regression based on the Mean Absolute Percentage Error(MAPE)measurement in determining the best method with accuracy of 5,4%. Increasing the inflation parameters of one unit will increase the IHSG value. But an increase in exchange rates and interest rates in one unit will reduce the IHSG value. Exchange rate has a significant effect on the IHSG using non-parametric regression. These predictions can be used by investors to consider a policy of holding shares or selling shares of a company.

Actions (login required)

View Item