Estimation of volatility in presence of high activity jumps and noise. Jean Jacod (UPMC-Paris 6)
We consider an It semimartingale which is observed along a discrete (regular or not) time grid, within a fixed time interval. The observations are contaminated by noise, and the semimartingale has jumps with a degree of activity bigger than 1. Our aim is to revisit the estimation of the integrated volatility in such a setting: we use a mixture of the pre-averaging method (to eliminate noise) and of the empirical characteristic function method, which has been shown to be effcient (after proper de-biasing) even when the jump activity is bigger than 1, in contrast with most other methods.
This talk is a presentation of a joint work with Viktor Todorov, from Northwestern University.