Title:Diffusive behavior and the modeling of characteristic times in limit order executions

Abstract: We present a study of the order book data of the London Stock Exchange for five highly liquid stocks traded during the calendar year 2002. Specifically, we study the first passage time (FPT) of order book prices needed to observe a prescribed price change "Delta", the time to fill (TTF) for executed limit orders and the time to cancel (TTC) for canceled ones. We find that the distribution of the first passage time decays asymptotically in time as a power law with an exponent "L_FPT ~ 1.5". The median of the same quantity scales as "Delta^1.6", which is different from the "Delta^2" behavior expected for Brownian motion. The quantities TTF, and TTC are also asymptotically power law distributed with exponents "L_TTF = 1.8-2.2" and "L_TTC = 1.9-2.4", respectively. For the medians of the time to fill we observe a scaling proportional to "Delta^1.4". We outline a simple model, which assumes that prices are characterized by the empirically observed distribution of the first passage time and orders are canceled randomly with lifetimes that are asymptotically power law distributed with a power "L_LT". The model predicts "L_TTF = L_TTC" and we estimate from empirical data "L_LT ~ 1.6". We verify that the observed results are not crucially dependent on simplifying assumptions.