Volatility and realized quadratic variation of differenced returns

This paper analyzes some asymptotic results for a new estimator of integrated volatility in a continuous-time diffusion process of high frequency data (used in asset pricing finance). The estimator, which is computationally efficient, is based on the quadratic variation of the second order log-price differences. This is contrary to the well known realized quadratic variation of intra daily returns (which is based on first order log-price differences). This latter is known as realized volatility. Analytically, the asymptotics of the proposed estimator is compared to the usual realized volatility estimators. Lastly, we provide some simulation experiments to illustrate the results.