Banco invest field lab on option volatility models - a detailed analysis of garch (P, Q) models -

Abstract

This project aims to analyze which volatility estimation model can better forecast volatility for Banco Invest. The compared models are the GARCH (1, 1), Exponentially Weighted Moving Average, Heston-Nandi GARCH, and two variations of the Heston stochastic volatility model. The model recommended for Banco Invest is the Heston, as it is the one that presents the closest results to the realized volatility and demonstrates the most stable estimates. Alternatively, if it is not Banco Invest’s intention to use the implied volatility as an in putw hen forecasting volatilities, the Heston-Nandi GARCH model should be taken in to consideration.