problem & doubts in forward testing.

New comment
 

I have been unclear on my doubts about forward testing even after researching a lot.


I test with a 1/3 forward testing period with criteria as "complex criterion max".  My optimization has 2835 passes.

This is the data for first optimization pass. --- After forward testing the optimization results table has 1191 rows while forward results table has 171 rows. Not all high values in the criteria value column(not the outstanding balance but the custom criteria value in orders of 0.x) of optimization results table are shown in forward results table.

The optimization results table has 178 profitable row and forward results has 38 profitable rows.

What does this data mean?

I have repeated this optimization for 6 cycles. But I don't understand this data.


1. How does this data reflect if the EA is bad or good. How to tell if an EA is good or bad by looking at this data.

2. Which set of data is to be passed to the forward period of last cycle i.e live trading period? I suppose it should be any one of the common values from the profitable rows in forward results tables from the previous set of forward testing cycles only. If that is so the data set to be passed should also checked to be common with a profitable result in the optimization results table from the last period of backtesting since its forward testing period will be the final period for live trading.

I have attached the screenshots for optimization & forward results table. Optimization Results Table Forward Results Table


try not to give any links since i have read most of the or perhaps all, but just guide me with detailed steps and reasoning.



please help. Ty

 

@Sergey Golubev 

@Marco vd Heijden

@Vladimir Karputov


Can you all help me pls?

 
Could you maybe rephrase your question?

I have tried to understand what you are trying to solve, but it stays somehow blurry.

Complex criteria needs to be evaluated by your own code.

Forward testing is used to verify your optimization runs.

Any EA or better said strategy is evaluated based on your own way of evaluation. There is no good or bad in an absolute scale. This depends very much on your way of applying.


 
Dominik Egert:
Could you maybe rephrase your question?

I have tried to understand what you are trying to solve, but it stays somehow blurry.

Complex criteria needs to be evaluated by your own code.

Forward testing is used to verify your optimization runs.

Any EA or better said strategy is evaluated based on your own way of evaluation. There is no good or bad in an absolute scale. This depends very much on your way of applying.


I'll rephrase it and ask again with new screenshots in few days.


Thanks.

New comment