In Support of Garch For Volatility Analysis and Forecasting

The study projects the misuse of linear regression for volatility analysis and forecasting. On testing the fundamental assumptions of linear regression in time series, Linear regression is not deemed fit for the use because of the violation of the modelprinciples. The Generalized Autoregressive Conditionally Heteroskedasticity (GARCH 1, 1) model is considered for estimating volatility. GARCH (1, 1) has proved to be an ideal model in the GARCH family as its more consistent and efficient with its results. A relationship has been derived between the NIFTY AND NIFTY F&O market movements. Therefore, spearman's correlation is performed on the prices of spot market index (NIFTY) and the volatility of the derivative market index (NIFTY F/O). Reasonablejustification for using spearman's correlation is because the variance data of the NIFTY F/O represents heavy positive skewness, which violated the fundamental assumption of the Pearson's correlation. The result of the correlation was significant. The tools used for the work are: linear regression, Durbin Watson test for autocorrelation, scatter plots, outliers labeling, normal distribution plots, regression plot, graphical representation and GARCH (1, 1) model. As a result a composite model has been framed with sound justification for estimating the volatility and movement impact of both the indexes.