Bias-corrected realized variance
提出一种基于不同采样频率下两个实现方差加权调整的偏差校正实现方差估计量，蒙特卡洛实验表明其有限样本方差显著减小，性能与常用积分方差估计量相当，适合研究者和从业者使用。
We propose a novel “bias-corrected realized variance” (BCRV) estimator based upon the appropriate re-weighting of two realized variances calculated at different sampling frequencies. Our bias-correction methodology is found to be extremely accurate, with the finite sample variance being significantly minimized. In our Monte Carlo experiments and a finite sample MSE comparison of alternative estimators, the performance of our straightforward BCRV estimator is shown to be comparable to other widely-used integrated variance estimators. Given its simplicity, our BCRV estimator is likely to appeal to researchers and practitioners alike for the estimation of integrated variance.