when we retest the strategies with the Montecarlo, SQX take all the sample. So, we have to test before the sample of the "in sample", delete the strategies not robust, and then make another retest with the sample of the "out of sample". It could be easier if we can choise the sample, if we can flag the montecarlo on "in sample", "out of sample", or both: we could view in one shot what strategies - with the montecarlo done on in sample - lose in out of sample.