Reply

Back-testing Performance

6 replies

beniko

Subscriber, bbp_participant, community, 31 replies.

Visit profile

8 years ago #114104

Hello Mark,

 

I notice that back-testing of EAs generated from EA wizard takes ages compared to conventional EAs.

 

What is the cause of that and is there a way to improve back-testing speed?

 

Regards,

Ben

0

Jason R

Subscriber, bbp_participant, community, 15 replies.

Visit profile

8 years ago #132200

Personally, The EAs I have generated with SQ run much faster than many of the others I have. I think it is entirely due to the complexity of the logic. 

 

I personally try to keep all logic that can be run on every bar open separate from anything that must be done tick-by-tick. then I try to use rules at the beginning of an IF sequence that will invalidate it as soon as possible.

 for example: If the logic is to check for conditions to exit a long position, I will use “if long” at the very beginning. that will invalidate the logic immediately whenever I’m flat. 

 

Just a noob myself, but my 2 cents worth.

I do know that coding isn’t just about making it work, but also about making it work efficiently. 

0

rjectweb

Customer, bbp_participant, community, 2 replies.

Visit profile

8 years ago #135255

Hi,

I think EA wizard is very good for creating EAs simply and fast, but I would also appreciate some tips to improve back-testing speed. My EAs are quite simple, but backtesting them takes considerable time, and for that reason it is difficult to try to optimize them.

 

I’m already using “Evaluate this rule only on bar opening”, but I was thinking in other measures like maybe removing some pieces of code from the mq4 file generated, or something. Any tips that may improve the backtesting speed could be of help. In the end, once an EA has been coded you will probably need to optimize it and try to find the better values for parameters and different pairs and timeframes, so in my opinion being able to backtest an EA fast is quite important.

 

Any ideas?

Thanks in advance

Regards

0

daveM

Subscriber, bbp_participant, community, customer, sq-ultimate, 110 replies.

Visit profile

8 years ago #135258

You might consider using less intensive testing at the start of optimization in order to get the parameters adjusted

0

rjectweb

Customer, bbp_participant, community, 2 replies.

Visit profile

8 years ago #135273

Yes, that can be done, thanks for answering, but this is a general rule for backtesting and not specific for EAs created with EA Wizard. Anyway, thanks for taking the time and trying to help.

 

That said, sorry if I insisit, but the main issue here is it seems (at least to the first poster and to me) that backtesting EAs created with the wizard is quite time consuming, more than EAs created manually by a mql4 coder. Does anybody know something that can be done specifically for EAs created with the wizard? If not, I recommend improving backtesting time as an improvement for future versions of EA wizard. I think that should be a big improvement for an already very good product.

 

Thanks in advance.

Regards

0

daveM

Subscriber, bbp_participant, community, customer, sq-ultimate, 110 replies.

Visit profile

8 years ago #135353

EA Wizard is coded so as to provide multiple shortcuts for the user…… meaning that under the hood is extensive coding that we do not see.

 

Processing time is consumed by the extra coding…..

0

Threshold

Customer, bbp_participant, community, 723 replies.

Visit profile

8 years ago #135395

Code strategies that enter/trail/stop out/ exit on bar open. Then use bar open only for backtest/opt. It will optimize several thousands iterations in just 1 hour.

SQ4 will be able to backtest/optimize EA Wizard strategies.

0

Viewing 6 replies - 1 through 6 (of 6 total)