A1232
Title: Range-based signed volatility estimation and the goodvol/bad vol conundrum
Authors: Massimiliano Caporin - University of Padova (Italy) [presenting]
Sandra Paterlini - University of Trento (Italy)
Pierdomenico Duttilo - University of Padova (Italy)
Abstract: The possibility of decomposing the estimate of the integrated variance into the contribution coming from positive (good) and negative (bad) returns has led to several empirical studies analyzing differences in persistence and interdependence between signed variance components. The signed decomposition is extended to range-based estimators of integrated variances, exploiting either the returns sign or the timing of minimum and maximum prices over a fixed intra-daily interval. By means of simulations, the convergence of the proposed estimators is addressed in various settings, including microstructure noise and jumps. Furthermore, the results are compared to classical realized variance estimators with real data, both at the descriptive level and within a forecasting exercise.
