MAXIMUM OVERLAPPING DISCRETE WAVELET METHODS FOR MODELLING THE SAUDI STOCK EXCHANGE

Abstract

This study forecasts the stock volatility based on wavelet-based generalized autoregressive conditional heteroscedasticity (GARCH) methods. It builds a forecast model based on GARCH methods, autoregressive integrated moving average (ARIMA) method, and maximum overlap discrete wavelet transform (MODWT) based on the best-localized function (Bl14) models. The aim is to measure the volatility of stock market forecasting through the non-linear spectral model, GARCH models, which are general GARCH (gGARCH), exponential generalized autoregressive conditional heteroscedasticity (EGARCH) and Glostsen-Jagannathan-Runkle GARCH (GJR-GARCH) functions, MODWT based on best-localized function (Bl14), and ARIMA model. Also, the study will build a prediction model based on GARCH, ARIMA and MODWT methods based on best-localized function models (Bl14). The developed model was used in the Saudi stock market from August 2011 to December 31, 2019. The study results show that the Saudi Stock Exchange Market witnessed high volatility in several periods. Market returns show a non-normal distribution indicating high volatility among returns. The highest closing price and return volatility was recorded in 2015 and 2016. The GARCH (1,1) model is the best model used to measure volatility. Instabilities are checked and displayed using MODWT based on B114 capabilities. The hybrid method is best for forecasting closing prices and returns in Tadawul Stock Exchange Market (TSEM). The study recommends that the GARCH model based on the normal distribution is the best for measuring volatility, and the hybrid method is the best method that can be used for forecasting.