Expecting the same parameters to work over a twelve year period, especially given what the markets did during this time, is a big ask.
Why don't you break it up into two year lots, so back test/optimise 2000/2001 and apply these results to 2002, then do 2001/2002 and apply these results to 2003 etc.