An update and a warning.............................
I have downloaded various hourly pair data from 1st July 2010 onwards
I have found the following......................
a) duplicate records
these include all of March 2012, 2013 and 2014
b) missing records
these include all of February 2011, 2012, 2013 and 2014
So, if you wish to have a clean data set - you will need to tidy up the Dukascopy data quite a bit
Bob M