Read here https://forexsb.com/forum/post/54210/#p54210

Another easiest method is to delete all the other in the list and keep only the Strategy you want. 

1. For example, delete all codes in SetSignals(Signal &signalList[]) but keep your codes line 683-685.  ie. Delete signal 000 to 011 and signal 013-099 so that only 012 is the only one remain.

2. Save and then compile the file

3. Go to EA Studio and use the import strategy back function. 

You should be able to see your single, selected Strategy.

Regards
Hannah

Sean, use your selected strategy (good strategy) and use it as a Preset to generate better ones.  This is an excellent method to improve your trading results.

To improve your strategy further, while using as a Preset, you can also varied the SL range to "tighten" it and force the software to find better and more accurate entries, i.e. eliminating false entries.

Regards
Hannah

Yes, start a new generator procedure with a good selected Strategy as Preset.

As for the Collection, it seem like most people just create lots of instances and let the Generator run for a long time and then collect the "final" Collection at the end of the hours.

I prefer to download the Collection file (and export it as Portfolio) at every few hours interval (the initial intervals shorter and as it goes further, the interval between each file saved is longer because the 1st few couple of 100 strategies are replaced faster in the initial process).

After a number of test/experiment, you will soon find the most optimal duration to run your Generator.  Running your Generator any further (beyond this optimal point is a waste of time and resources).

Why do I prefer to collect different interval Collection for each instance?

Curve Fitting is the issue.  We all know as the Generator  runs longer, it's simply finding the best fit that has the highest yield (performance metrics).  So if you are only saving the last Collection file after many hours of running the Generator, the strategies you are getting are most likely to be more curve fitted than those earlier strategies.

So you can do an experiment to test out whether strategies in earlier Collection do better/worst than strategies save in later Collections.

1) Save the Collections in several intervals (for the sake of the experiment, use a short and reasonable time intervals)

2) Demo test your Collections results and see which Collection files yield better results.

From my experiments, it varied (due to the intervals I use).  Usually it's the collections found in the middle.  Hardly any good ones (relatively speaking) at the last collections. 

Take Note: The result varied due to many circumstances, I'm not saying "ALL last collections" are bad.  I'm asking people to test out instead of assuming that the Last Collection is the Best.  If people agree that the common problem in strategy search is the issue of Curve Fitting then it is important to understand how long you run your instances will result to increasing the effects of Curve fitting issues occurring in your collections.

Regards
Hannah