UPD:
– for the successful training bases created, I will provide an adviser for temporary use for free.
– the training bases will be laid out as you learn.
– training requires about 20 epochs. 

It is possible to use one of two strategies – either trade in 2 directions, or use SL. When using SL, the trading results will be identical to the learning results. Since only 1 order is used at a time during training. But! – the ability to trade both ways (with SL =0)  sometimes it allows you to get out of a difficult situation without losses.

Because the ADVISER is resource-intensive and the market is not able to process it, the TypeOfWork parameter with the Market value is entered.

It must be switched to any other desired value! 

Published for co-education!

The depth of the input data set is 50 bars of the timeframe specified in the settings.

ThresholdOUT has no effect.

speed in some cases affects only at a very large value.

In the training mode, we open only 1 order with a minimum lot with equal SL and TP. The schedule should be uniform. In this mode, the profit itself does not matter.

Distance only works for MaxOrders>1. >MaxOrders>1> slows down a lot.

LearnEpoch is an unused parameter for organizing cycles in optimization.

[spoiler title=”Read More…”]

If you delete the data_w1_[TF]_[INSTRUMENT].csv and data_w2_[TF]_[INSTRUMENT].csv files or change PeriodBars in the parameters, the training will start over.

Location of C files:Users[USER]AppDataRoamingMetaQuotesTerminalCommonNeuro

[/spoiler]