Undergraduate Honors Thesis Presentation: Bootstrap estimation for pre-averaged realized volatility under market microstructure noise

Abstract: Volatility estimation is a big topic in financial econometrics, which would be useful for asset pricing or market analysis and many other areas. We developed and applied a new bootstrap method to stock market data and achieved an unbiased estimation of the volatility of the diffusion process under a high-frequency dataset, innovatively combining wild bootstrap and blocks of blocks bootstrap methods. Contrary to the traditional assumption that the market is frictionless, microstructure noise in the market, which would cause severe bias as the frequency increases, was considered. As the pre-average method used for bias removal contains finite sample distortions if the frequency is not high enough, we further took the resampling method as a bootstrap to make a more consistent and unbiased estimation and inferences.

Host: Jose Figueroa-Lopez