Evaluating forecast accuracy for NSE Nifty: A cross-market study using OLS, HSC, GARCH, and VAR models

Abstract

An investigation is conducted into the predictive and interconnected behaviour of NSE NIFTY in relation to major global stock indices over a twenty-five-year period, using data obtained from reputed international financial databases. The analysis incorporates four complementary econometric frameworks Ordinary Least Squares (OLS), Heteroskedasticity and Serial Correlation Consistent estimation (HSC), the GARCH(1,1) volatility model and the Vector Autoregressive (VAR) system to capture linear relationships, robust short-run linkages, time-varying volatility patterns, and multidirectional spillover effects. The indices are organised into regional and developmental groups to examine differential transmission mechanisms across advanced, Asian, and emerging market environments. The empirical evidence indicates that Asian and emerging markets exert consistently strong and significant influence on NSE NIFTY, while several European indices display weaker or mixed predictive contributions. HSC offers the most precise parameter estimates and favourable information criteria, VAR effectively captures dynamic inter-market interactions and GARCH provides improved modelling of volatility clustering, especially during high-uncertainty periods. Overall, predictive performance strengthens as a wider set of international markets is incorporated, highlighting the increasing global integration of Indian equities and the importance of combining linear, robust, and volatility-sensitive frameworks for comprehensive market forecasting.