A jump and noise robust kernel based estimator of quarticity for high frequency data

Integrated quarticity and its estimation has been a topic of interest in recent high-frequency financial econometrics literature. As it refers to the fourth power of the volatility, it is used in modelling and estimating the variability of the realized volatility of an asset. The existing estimators of quarticity fail to account for any jumps or microstructure noise during estimation. Hence in this paper we establish a kernel based robust estimator of quarticity which outperforms other standard estimators under these special scenarios. Extensive simulation and sensitivity analyses is carried out to showcase the application and comparison of the proposed estimator. The estimator is applied to find the quarticity among high frequency returns of five of the top cryptocurrencies.