Abstract: Volatility is a hottopic in financial research.People pay more and more attention to the high frequency data in finance because it contains more volatility information of intraday return than low frequency data does.Realized volatility is a completely new method to calculate volatility of high frequency data, which is applied widely in the study of high frequency data in finance.There are many methods to improve on the realized volatility for it has shortcomings of bigerror and non-robust ness.Among these,only the realized bipower variation overcomes the shortcoming of non-robust ness.The concept of weighted realized bipower variation which is put forward in this article,is not only robust but also unbiased and efficient. The theorem poof and the demonstration study also show the same conclusion:it can measure the volatility more precisely.

Key words: realized bipower variation, weighted realized bipower variation, realized volatility, efficiency, robust ness