The tick structure of the financial markets entails discreteness of stock price changes. Based on this empirical evidence, we develop a multivariate model for discrete price changes featuring a mechanism to account for the large share of zero returns at high frequency. We assume that the observed price changes are independent conditional on the realization of two hidden Markov chains determining the dynamics and the distribution of the multivariate time series at hand. We study the properties of the model, which is a dynamic mixture of zero-inflated Skellam distributions. We develop an expectation-maximization algorithm with closed-form M-step that allows us to estimate the model by maximum likelihood. In the empirical application, we study the joint distribution of the price changes of a number of assets traded on NYSE. Particular focus is dedicated to the assessment of the quality of univariate and multivariate density forecasts, and of the precision of the predictions of moments like volatility and correlations. Finally, we look at the predictability of price staleness and its determinants in relation to the trading activity on the financial markets. Copyright © 2021 Informa UK Limited

Dynamic Discrete Mixtures for High-Frequency Prices / Catania, Leopoldo; Di Mari, Roberto; Santucci de Magistris, Paolo. - In: JOURNAL OF BUSINESS & ECONOMIC STATISTICS. - ISSN 0735-0015. - 40:2(2022), pp. 559-577. [10.1080/07350015.2020.1840994]

Dynamic Discrete Mixtures for High-Frequency Prices

Paolo Santucci de Magistris
2022

Abstract

The tick structure of the financial markets entails discreteness of stock price changes. Based on this empirical evidence, we develop a multivariate model for discrete price changes featuring a mechanism to account for the large share of zero returns at high frequency. We assume that the observed price changes are independent conditional on the realization of two hidden Markov chains determining the dynamics and the distribution of the multivariate time series at hand. We study the properties of the model, which is a dynamic mixture of zero-inflated Skellam distributions. We develop an expectation-maximization algorithm with closed-form M-step that allows us to estimate the model by maximum likelihood. In the empirical application, we study the joint distribution of the price changes of a number of assets traded on NYSE. Particular focus is dedicated to the assessment of the quality of univariate and multivariate density forecasts, and of the precision of the predictions of moments like volatility and correlations. Finally, we look at the predictability of price staleness and its determinants in relation to the trading activity on the financial markets. Copyright © 2021 Informa UK Limited
2022
Dynamic mixtures; EM Algorithm; High-frequency prices; Skellam distribution; Volatility; Zeros.
Dynamic Discrete Mixtures for High-Frequency Prices / Catania, Leopoldo; Di Mari, Roberto; Santucci de Magistris, Paolo. - In: JOURNAL OF BUSINESS & ECONOMIC STATISTICS. - ISSN 0735-0015. - 40:2(2022), pp. 559-577. [10.1080/07350015.2020.1840994]
File in questo prodotto:
File Dimensione Formato  
JBES_2020.pdf

Solo gestori archivio

Tipologia: Documento in Post-print
Licenza: DRM (Digital rights management) non definiti
Dimensione 4.09 MB
Formato Adobe PDF
4.09 MB Adobe PDF   Visualizza/Apri
JBES_2022.pdf

Solo gestori archivio

Tipologia: Versione dell'editore
Licenza: Tutti i diritti riservati
Dimensione 3.53 MB
Formato Adobe PDF
3.53 MB Adobe PDF   Visualizza/Apri
Pubblicazioni consigliate

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/11385/199237
Citazioni
  • Scopus 7
  • ???jsp.display-item.citation.isi??? 8
social impact