Use this URL to cite or link to this record in EThOS: https://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.759716
Title: Robust volatility estimation for multiscale diffusions with zero quadratic variation
Author: Manikas, Theodoros
ISNI:       0000 0004 7431 7457
Awarding Body: University of Warwick
Current Institution: University of Warwick
Date of Award: 2018
Availability of Full Text:
Access from EThOS:
Access from Institution:
Abstract:
This thesis is concerned with the problem of volatility estimation in the context of multiscale diffusions. In particular, we consider data that exhibit two widely separated time scales. Fast/slow systems of SDEs that adopt a homogenized SDE are employed to model such data. The problem that one is confronted with, is the mismatch between the multiscale data and the homogenized SDE. In this context, we examine whether if by using the multiscale data, the diffusion coefficient of the homogenized SDE can be estimated. Our proposed estimator consists on subsampling the initial data by considering only the local extremals to overcome the issue associated with the underlying model. We provide both theoretical and numerical heuristics, suggesting that our proposed estimator when it is applied to multiscale data of bounded variation is asymptotically unbiased for the volatility coefficient of the homogenized SDE. Furthermore, for the particular example of a multiscale Ornstein-Uhlenbeck process, the numerical results indicate that the L2-error of our estimator is very small. Moreover, we illustrate situations where the proposed estimator can also be used for multiscale data with bounded non-zero quadratic variation.
Supervisor: Not available Sponsor: University of Warwick
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID: uk.bl.ethos.759716  DOI: Not available
Keywords: QA Mathematics
Share: