I've tried a few time slices and the example I posted is typical.  The BCW site shows numbers that are not in the API trade data.  However I have not checked it against other time intervals.  Perhaps a 3min interval posted at 9:45 does not mean 9:45 to 9:42 but something else.  Since I've run into a lot of weird problems with the btc-e trade data I thought someone here might be familiar with it or BCW methods.  Maybe I should run it over 1 min intervals to reduce any errors in how the time frames are defined.

I'm using python with a sqlite3 database and matplotlib modules.