The present thesis is focused on the application of the Realized Peak-over-Threshold method proposed by Bee, Depuis, and Trapin (2019) [12], which aims to enhance the traditional model by using time-dependent covariates derived from the information contained in high-frequency data, widely available nowadays. The results obtained from the in-sample fitting on the MRK stock show that high-frequency data contributes to modeling the behaviour of loss extremes, providing additional information compared to that contained in daily data. Furthermore, the out-of-sample Value at Risk (VaR) prediction achieved with high-frequency measures performs better than the one traditionally obtained with low-frequency measures.
Il presente lavoro di tesi è focalizzato sull’applicazione del metodo Realized Peak-over-Threshold proposto da Bee, Depuis e Trapin (2019) [12] che si ripropone di arricchire il modello tradizionale attraverso l’utilizzo di covariate tempo-dipendenti ricavate a partire dall’informazione contenuta nei dati ad alta frequenza, oggigiorno ampiamente diffusi e disponibili. I risultati ottenuti dall’adattamento in-sample sul titolo MRK mostrano che l’alta frequenza contribuisce a modellare il comportamento degli eccessi di perdite, aggiungendo informazione rispetto a quella contenuta nei dati giornalieri. Inoltre, la previsione out-of-sample del VaR ottenuta con le misure di alta frequenza performa meglio di quella ottenuta tradizionalmente con misure di bassa frequenza.
Teoria dei Valori Estremi con Dati ad Alta Frequenza: Applicazione del Metodo Realized POT al titolo MRK
SALIERNO, MARTINA
2023/2024
Abstract
The present thesis is focused on the application of the Realized Peak-over-Threshold method proposed by Bee, Depuis, and Trapin (2019) [12], which aims to enhance the traditional model by using time-dependent covariates derived from the information contained in high-frequency data, widely available nowadays. The results obtained from the in-sample fitting on the MRK stock show that high-frequency data contributes to modeling the behaviour of loss extremes, providing additional information compared to that contained in daily data. Furthermore, the out-of-sample Value at Risk (VaR) prediction achieved with high-frequency measures performs better than the one traditionally obtained with low-frequency measures.File | Dimensione | Formato | |
---|---|---|---|
Salierno_Martina.pdf
accesso aperto
Dimensione
2.92 MB
Formato
Adobe PDF
|
2.92 MB | Adobe PDF | Visualizza/Apri |
The text of this website © Università degli studi di Padova. Full Text are published under a non-exclusive license. Metadata are under a CC0 License
https://hdl.handle.net/20.500.12608/64211