For those users who would like to understand how we computed the "Oxford-Man Institute's realised library" we have prepared some details.

The raw high frequency data which powers our realized library was downloaded by us from the  Reuters DataScope Tick History database. We refer to that website for details of this data.  The documentation given here is split into two parts

  1. data cleaning: including conventions about holidays and start of the day
  2. computing the econometric statistics: including bandwidth selection or window width

Each of these parts is self-contained.

Code for computing data cleaning and computing econometric statistics are given in Matlab and OxMetrics here.

The documentation includes two days of uncleaned data, code for cleaning it which produces two days of cleaned data.  Then we also give our code to compute our statistics for those two days.  These numbers are then the ones which appear in our library.