Preprint B56/2013
Turbulência e Acoplagem como Medidores de Risco
Gustavo Rodriguez Pe├žanha
Keywords: Risk | Crisis | Turbulence
Risk modelling, specially crisis, is still a challenge in financial studies. The present work is based on an approach similar to the one used by Mark Kritzman, using two measures: coupling and turbulence. The first one is based on principal components analysis and the second one on Mahalanobis distance. The latter shows how far from its mean a group of returns is, taking into consideration correlation among assets. It is shown that the coupling and turbulence metric results are sometimes better, sometimes worse than the Exponentially Weighted Moving Average (EWMA), but when used together a better result is achieved regarding crisis prediction, according to the chosen criteria.