Abstract
Com base em estudos desenvolvidos em anos recentes sobre o uso de dados de alta frequência para a estimação da volatilidade, este artigo implementa o modelo Autorregressivo Heterogêneo (HAR)desenvolvido por Andersen, Bollerslev, e Diebold (2007) e Corsi (2009), e o modelo Componente (2-Comp) desenvolvido por Maheu e McCurdy (2007) e os compara com a família de modelos Autorregressivos com Heteroscedasticidade Generalizados (GARCH)para estimar a volatilidade e os retornos. Durante o período analisado, os modelos que usam dados intraday obtiveram melhores previsões de retornos dos ativos avaliados, tanto dentro como fora da amostra, confirmando assim que esses modelos possuem informações importantes para uma série de agentes econômicos.
Highlights
High-frequency data is the result of observations made available over short periods of time
Among other seminal articles which deal with the unique properties and characteristics of the distribution of intraday data, it is possible to quote: Zhou (1996), who used ultra-high-frequency data relevant to the currency exchange markets in order to explain the negative autocorrelation of the first order of returns and to estimate volatility for high-frequency data, Goodhart and O’Hara (1997) which highlight the effects of market structure on the interpretation and analysis of the data, the effects of intra-day seasonal and the effects of time-varying volatility, and Andersen and Bollerslev (1997, 1998a) who analyzed the behavior of intraday volatility, the volatility shocks due to macroeconomic pronouncements and the long-term persistence in the temporal series of realized volatility, on the currency exchange market
The most noteworthy article in the Brazilian market is that developed by Wink Junior and Valls Pereira (2012) which, in a pioneering manner, choose the optimal intraday time interval, deal with the question of noise generated by the market microstructure and implement two recent models, which use high-frequency data to estimate and forecast the volatility of five representative shares of the Bovespa Index
Summary
High-frequency data is the result of observations made available over short periods of time. The availability of trader data-bases and the calculation advances have made this data increasingly accessible to researchers and traders and have generated an enormous growth in the empirical research in finance This development has opened the way for a vast array of empirical applications, in particular on liquid financial markets, dealing large volumes and frequency of negotiations and low transaction costs. Models implemented can be used to validate and to refine intraday price and return models They can be useful in intraday investment strategies, in long-short strategies and in risk management, for instance to calculate different conditional volatilities in order to compare and to improve Value at Risk methodologies.
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
