When I let Evolver run for a night it tends to produce results like this:
All the top ranks are occupied by very similar strategies. It does not help to let it run further as the probability for a new star, better then the existing ones is very, very low.
I.e. the picture will not change significantly during the next hours.
When I restart Evolver for the next night it will produce a similar picture but with a completely different type of strategy.
This recurring observations sparked another #FeatureRequest:
1. Detect if Evolver has converged to a stable type of strategy. This might be the case if the first 5 or 10 strategies are very similar and if the target metric does not improve significantly for some time.
2. As soon as the Evolver has reached convergence stop it.
3. Save the top strategy in a special folder and generate an entry in a special report.
4. Restart the Evolver.
I think this procedure is more efficient in finding new ideas for strategies than letting the Evolver run for a long time.
If this text makes sense to you and you are like me and want to discover new exciting trading ideas, then please vote for this #FeatureRequest.
All the top ranks are occupied by very similar strategies. It does not help to let it run further as the probability for a new star, better then the existing ones is very, very low.
I.e. the picture will not change significantly during the next hours.
When I restart Evolver for the next night it will produce a similar picture but with a completely different type of strategy.
This recurring observations sparked another #FeatureRequest:
1. Detect if Evolver has converged to a stable type of strategy. This might be the case if the first 5 or 10 strategies are very similar and if the target metric does not improve significantly for some time.
2. As soon as the Evolver has reached convergence stop it.
3. Save the top strategy in a special folder and generate an entry in a special report.
4. Restart the Evolver.
I think this procedure is more efficient in finding new ideas for strategies than letting the Evolver run for a long time.
If this text makes sense to you and you are like me and want to discover new exciting trading ideas, then please vote for this #FeatureRequest.
Rename
I'm in! (Dr. K, make sure you vote for it too!)
OK with such overwhelming support I guess this is the next item on the docket for Build 35! I think we'll call the feature "Apex Strategies."
>3. Save the top strategy in a special folder and generate an entry in a special report.
If there are several strategies with the same profit, my suggestion is to choose the one with the least criteria in blocks, because now there are many similar ones, which add extra parameters that do not affect anything
If there are several strategies with the same profit, my suggestion is to choose the one with the least criteria in blocks, because now there are many similar ones, which add extra parameters that do not affect anything
Your Response
Post
Edit Post
Login is required