High-frequency volatility modelling:a Markov-switching autoregressive conditional intensity model

We develop a Markov-Switching Autoregressive Conditional Intensity model for high-frequency volatility modelling via the absolute price change point process. By incorporating a regime-switching relationship between price durations and trading volume, we discover two distinct regimes with a dominant regime exhibiting a strong correlation between price durations and trading volumes, and a minor regime showing a much weaker correlation. Observations of the dominant regime spread evenly across trading days while those of the minor regime cluster around the start and end of trading days. These findings suggest that the minor regime represents the information arrival into the market due to its appearance with the well-documented diurnal pattern of information arrival, whereas the dominant regime corresponds to the volatility associated with low information content trading. We provide a high-frequency measure of the information content in the market, and a measure of the impact of information on the volatility process.