Regime switching model for financial data: empirical risk analysis - Inria - Institut national de recherche en sciences et technologies du numérique Accéder directement au contenu
Article Dans Une Revue Physica A: Statistical Mechanics and its Applications Année : 2016

Regime switching model for financial data: empirical risk analysis

Résumé

This paper constructs a regime switching model for the univariate Value-at-Risk estimation. Extreme value theory (EVT) and hidden Markov models (HMM) are combined to estimate a hybrid model that takes volatility clustering into account. In the first stage, HMM is used to classify data in crisis and steady periods, while in thesecond stage, EVT is applied to the previously classified data to rub out the delay between regime switching and their detection. This new model is applied to prices of numerous stocks exchanged on NYSE Euronext Paris over the period 2001-2011. We focus on daily returns for which calibration has to be done on a small dataset. The relative performance of the regime switching model is benchmarked against other well-known modeling techniques, such as stable, power laws and GARCH models.The empirical results show that the regime switching model increases predictive performance of financial forecasting according to the number of violations and tail-loss tests. This suggests that the regime switching model is a robust forecasting variant of power laws model while remaining practical to implement the VaR measurement.
Fichier principal
Vignette du fichier
estimation-power-law.pdf (680.8 Ko) Télécharger le fichier
Origine : Fichiers produits par l'(les) auteur(s)

Dates et versions

hal-01095299 , version 1 (11-02-2015)
hal-01095299 , version 2 (02-06-2016)

Identifiants

  • HAL Id : hal-01095299 , version 1

Citer

Khaled Salhi, Madalina Deaconu, Antoine Lejay, Nicolas Champagnat, Nicolas Navet. Regime switching model for financial data: empirical risk analysis. Physica A: Statistical Mechanics and its Applications, 2016. ⟨hal-01095299v1⟩
7036 Consultations
2459 Téléchargements

Partager

Gmail Facebook X LinkedIn More