Order Flow, Transaction Clock, and Normality of Asset Returns: A Comment on Ané and Geman (2000)

We investigate the procedure used by Ané and Geman (2000) to recover the moments of information flow from high frequency data in a model which generalizes the subordinated / mixture of distributions process in Clark (1973). Using Monte Carlo experiments we show that the third and higher moments of the latent information flow cannot be accurately recovered using their univariate procedure. We explain why this happens. In our data, returns conditioned on the recentered number of trades are not Gaussian.