Some of us like to build on selected time frame only precision then run a crosscheck higher precision retest in the same build task. It works OK using minimal resources till the main data filters are passed then the crosscheck happens. The false positives (strategies that look good on selected only precision but do bad on higher precisions) don't make it in to the RESULTS tab because of the higher precision filters, that's good,
but they stay in the gen pool getting all the "love" (higher likelihood of surviving and procreating and mutating) from the gen algo. It would be nice if the retest filters actually removed the strategy from the gen pools as the text implies "dismissed (thrown away)" in the filter setting notes.
OR - we could rank fitness via the higher precision test to avoid this too. There is one more important point about that though. The crosscheck retest higher precision should NOT run unless the main data filters have passed. So fitness should rank by main data unless they pass the main data filters and have been subject to the retest in which case rank by retest. I bring this up because when rank by portfolio is selected then the retest on other markets happens on all strategies upon the first generation regardless of the main data filters. That is all strategies that pass init pop filters are retested automatically when rank by portfolio is selected and I don't think we want this for rank by retest higher precision.