One of the stylized facts of the financial markets states that the log-returns of the stock price on consecutive daysri= log(Si/Si1) are (pick one):

i) Positively correlated

ii) Negatively correlated

iii) Uncorrelated

1) Does this property hold also for the squared returnsri2and the absolute

values of the returns|ri|?

2) Does this stylized fact remain valid for microstructure data? If not,

what is different at the microscale?