Some Estimation Methods in High Frequency Financial Data

Time: 15:00 to  16:30 Ngày 29/03/2019

Venue/Location: C2-714, VIASM

Speaker: Dr. Rituparna Sen - Indian Statistical Institute, India

Content:

First, using diffusion processes driven by fractional Brownian motion (fBm) to model stock prices, a central limit theorem is derived for the quadratic variation as an estimator for volatility for constant as well as time varying volatility. Next, the problem of modelling bivariate intraday financial data through Copula is discussed when prices are observed non-synchronously. We show that this may lead to serious underestimation of the Copula parameter and propose a modification to make the estimator unbiased. Finally, the limiting spectral distribution of high dimensional covariance matrix is derived for non-synchronous financial data. We  performed simulations to test the validity of our result and applications to market data.