Implementazione della tecnica AFO per la stima dei parametri di un modello GARCH(1,1). Analisi di robustezza e confronto prestazionale con i solver tradizionali

In the calculation of theoretical price for derivatives, one of the most critical parameter to estimate is the volatility of the underlying: a wrong choice of the model or an approximate estimation through maximum likelihood method leads to significantly different fair values. This paper mainly regards on the importance of implementing an efficient and reliable solver which allows to accurately estimate the parameters of one of the most widespread econometric model: GARCH(1,1). The study therefore highlights the critical issues that can be found in the usual optimization routines and how these can be solved thanks to the implementation of a robust deterministic global search technique such as the “Attraction ForceOptimization” – AFO.