An empirical investigation of Australian Stock Exchange data

We present an empirical study of high frequency Australian equity data examining the behaviour of distribution tails and the existence of long memory. A method is presented allowing us to deal with Australian Stock Exchange data by splitting it into two separate data series representing an intraday and overnight component. Power-law exponents for the empirical density functions are estimated and compared with results from other studies. Using the autocorrelation and variance plots we find there to be a strong indication of long-memory type behaviour in the absolute return, volume and transaction frequency.