With the availability of high-frequency data ex post daily (or lower frequency) nonparametric volatility measures have been developed, that are more precise than conventionally used volatility estimators, such as squared or absolute daily returns. The consistency of these estimators hinges on increasingly finer sampled high-frequency returns. In practice, however, the prices recorded at the very high frequency are contaminated by market microstructure noise. We provide a theoretical review and comparison of high-frequency based volatility estimators and the impact of different types of noise. In doing so we pay special focus on volatility estimators that explore different facets of high-frequency data, such as the price range, return quantiles or durations between specific levels of price changes.The various volatility estimators are applied to transaction and quotes data of the S&P500 E-mini and of one stock of Microsoft using different sampling frequencies and schemes. We further discuss potential sources of the market microstructure noise and test for its type and magnitude. Moreover, due to the volume of high-frequency financial data we focus also on computational aspects, such as data storage and retrieval.
Dieser Eintrag ist Teil der Universitätsbibliographie.
Search Authors in