Modelling and forecasting intraday volatility with range-based GARCH models for Indian stock market

This paper focuses on the range-based volatility estimations with standard GARCH, EGARCH and TGARCH models to analyse how the performance of the models changes in different range-based analysis. The idea is to check the suitability of these models on a unique 7 months 30-minute intraday Nifty data with different measures of range-based volatility measures such as Parkinson's, Garman-Klass, Garman Klass-Yang Zhang and Yang Zhang for their appropriateness on the Indian stock market. Robust empirical analysis and comparisons give two significant performance-based results: 1) range-based volatility forecasting models appear to perform better as compared to the return-based models; 2) asymmetric GARCH models are found to be a better fit for the Indian stock market vis-à-vis the basic GARCH model. Another major outcome shows that the GKYZ volatility estimator comes out to be the best for accurately forecasting the current period volatility of the Indian stock market with having the minimum forecast error. Also, the positive size test significance suggests that the magnitude of the good news has a larger impact on volatility as compared to the impact of the magnitude of the bad news for Nifty 50.