Quote from infolode:
You retrain these every 5-7 days? On the same instrument, data set and time frame?
Seems like you'd get Stop Polishing dialogs when Error less than:0.0??? This tells the net when to give up on a particular neuron. When the performance gains are less than 0.001%, for example, there is no need to continue searching for better solutions, as any gains will be practically insignificant.
Just curious because I train and tweek mine quite often too.
Custom scripts is a freakin' nightmare to initally set-up and deploy but worth the effort in the long run.
Garbage and generated outliers......alotta grey hair
![]()
Yes, but more like every 10 trading days or due to a significant change in stat-vol. I continue running the last-gen as well.

