CMStatistics 2017: Start Registration
View Submission - CFE
Title: Volume, durations and jumps in SV models for the evolution of intraday financial volatility Authors:  Antonio Santos - University of Coimbra (Portugal) [presenting]
Abstract: The use of stochastic volatility (SV) models in the characterisation of the intraday volatility evolution is analyzed. The novelty is the introduction of a set of information elements that can be retrieved from intraday data, like volume of transactions and durations. This kind of information allows also the consideration of jumps in returns and respective volatility, which is done through a subordinated duration stochastic process. A bivariate state-space model relates returns and durations with a latent state. The probabilities of the jumps are modelled through a process using a logit kind of relationship, which allows time-varying probabilities for the jumps. Volume-domain returns are used, which when mapped to the time-domain, represent unequal time-spaced observations, and the duration between the successive volume-domain returns can be calculated. This kind of variable may give information on the evolution of the volatility defined through a state-space model, and can be related to the probability of a jump in returns or volatility. Our approach is tested using 2 years of intraday data associated with 10 stocks traded in US markets.