Estimating and Forecasting Volatility Using Arima Model : A Study on NSE, India

Volatility had been used as an indirect means for predicting risk accompanied with the asset. Volatility explains the variations in returns. Forecasting volatility had been a stimulating problem in the financial systems. The study examined the different volatility estimators and determined the efficient volatility estimator. The study described the accuracy of forecasting technique with respect to various volatility estimators. The methodology of volatility estimation includes Close, Garman-Klass, Parkinson, Roger-Satchell and Yang-Zhang methods and forecasting is done through ARIMA technique. The study evaluated the efficiency and bias of various volatility estimators. The comparative analyses based on various error measuring parameters like ME, RMSE, MAE, MPE, MAPE, MASE, ACF1 gave the accuracy of forecasting with the best volatility estimator. Out of five volatility estimators analysed over a period of 10 years and critically examined for forecasting volatility, the research obtained Parkinson estimator as the most efficient volatility estimator. Based on various error measuring parameters, Parkinson estimator had been examined as more accurate estimator than any other estimator based on RMSE, MPE and MASE in forecasting through ARIMA Technique. The study suggests that the forecasted values had been accurate based on the values of MAE and RMSE. This research was conducted in order to meet out the demand of knowing the efficient volatility estimator for forecasting volatility with high accuracy by the traders, option practitioners and various players of stock market