First I will allow FSB Pro generator to breed 100 strategies in a collection, using 50000 bars of historical data. Then I will stop the generator and enlarge historical data period to 100000 bars. Now I will double click on a strategy in the collection to reload it for checking if it will show similar positive behavior in large historical data set. If not, I will delete this strategy from the collection, and proceed with next strategy. At the end I will arrive at the resulting few strategies of better quality, to check those more thoroughly.

During this process I discovered that amount of memory used in this process will grow steadily, until it will reach maximum for the system (in my case it is 16GB). I am using Windows Task Manager to show the amount of physical memory used. When reaching the limit of maximal system memory (which will make the system and FSB Pro work in slow motion), I have to stop the FSB Pro (first, of course, saving the collection) and restart FSB Pro again. I will then reload the collection and continue with sorting of the strategies in the collection (FSB Pro will start the process with about 2 GB of memory). I have to do this process of reloading FSB Pro two times to check all of 100 strategies in a collection.

I am not a professional programmer, but I suspect that some of the resources used by FSB Pro are not returned back to the system after the use.