Author's - page 10

[Serj]

hrenfx:

Thank you, that makes sense now:

1. The training is based on patterns calculated for bars_future bars before the current moment.
2. The pattern for the current moment is compared with the trained one.

The idea is much better than originally imagined.

Yes, everything is correct. But I am not satisfied with the results. I am more inclined to the variant with fixed stops. Maybe I will add signals from other instruments as well.

[Serj]

Sometimes there are good results, after a vertical forward streak.

But trades are few, apparently there are very few stable patterns. Or one may try it on smaller frames.

Well, I will have to analyze and test it.

[hrenfx]

her.human:

Unfortunately, due to personal circumstances, I cannot continue the conversation with you and others on this thread. Thank you for your constructive input.