Well at least for me, i dont run anything thats not looking OK on the whole backtest.

If you got like 40-60% looking good but rest is shit, thats not valid for me, that screams overoptimization (obviously, i would say!)  and the risk of that system breaking is way higher.

Ive posted several screenshots of my systems (again, just me and my opinions as usual), heres one strategy tested in Wall st, US tech, Us 500

Edit: I really wanna put emphasis on the fact that even though a system has 40% bad and 60% good periods in the backtest, it might continue to work going forward! BUT and theres a big buttt here.. From my own experience, and from what ive read and listened to, The longer “Good backtest” you got, the more chance there is that the system will keep on living and performing.