For more information about this meeting, contact Manfred Denker, Jason Morton.
|Title:||The Estimation of Spot Volatility for High Frequency Data|
|Seminar:||Seminar on Probability and its Application|
|Speaker:||Axel Munk, University of Goettingen|
|In this talk estimation of the pathwise (spot) volatility is discussed for
the price process of financial instruments which are traded at a high
frequency. It is known that on small time scales classical (nonparametric)
large scale approaches fail.
Microstructure noise is one way to explain this and to model the small time
In the first part of this talk we discuss Fourier series estimators for the
problem of estimating
a deterministic spot volatility in a microstructure noise model.
We show that these obtain optimal convergence rates over Sobolev ellipsoids.
In the second part we extend this to stochastic spot volatility. Due to the
unknown smoothness of the spot volatility it becomes necessary to introduce
locally adaptive estimators. To this end a specific thresholding for wavelets
based on preaveraged observations will be used. The resulting estimator can
be shown to be adaptive over a large range of Besov bodies. Our methods will
be used to analyze price processes on small temporal scales and we show that
various scaling paradoxes can be resolved.
This is joint work with Marc Hoffmann (Paris IV) and Johannes Schmidt-Hieber
Room Reservation Information
|Date:||12 / 10 / 2010|
|Time:||02:30pm - 03:25pm|