As i got it, the Montercarlo test goes to make modifcations about historical data, execution problems and others parameters.
Now i'm wondering if this test must be done on the "in sample data" or in "out of the sample data" in order to avoid over-fitting situations.
In addition, does the Montecarlo simulation be a substitute of the "out of sample" test?
Pratically, i would like to know which kind of connection exist between these two tools.

Thanks