Forecasting LQ45 Index Stock Price Using Hybrid ARIMA-GARCH Model and Walk Forward Validation
Dian Lestari- Arman Haqqi Anna Zili- Sindy Devila- Derick Hendri

Universitas Indonesia

Abstract

Modeling and forecasting the stock price are an important thing for investors. The stock price usually has high volatility because it always fluctuate as time goes and these changes vary from one point of time to another. Many models have been created to modeling and forecasting the stock price in several previous studies. In this study, the Hybrid ARIMA-GARCH model will be used to forecast the stock price. This hybrid ARIMA-GARCH model is used because ARIMA model is unable to handle high volatility data. The data used in this study is the daily closing price of two stocks that is part of the LQ45 index Stock. Those data will be split into train and test data with proportion 4:1. The train data is used to fit the model and get the forecasted stock price at the next day. The forecast of next several day will be found using a process named Walk Forward Validation. Error of this forecast is found by comparing the forecast results with the test data and calculated using MAE and RMSE. After calculate the error, it is found that the Hybrid of ARIMA (1,1,1)-GARCH (1,1) yield the best forecast of the data which means this model yield a forecast with minimum error.

Keywords: forecast- model- stock price- ARIMA-GARCH- Walk Forward Validation

Topic: Mathematics