Optimization of Optimization and Backtest
Please consider which section is most appropriate — https://www.mql5.com/en/forum/172166/page6#comment_49114893
I run some tests with an EA to find out how run time for EA changes for equal count of deals.
EA runs on M1 time frame and there are thousand of deals for less than 3 months. I think number of data is good enough to consider bias as non-effective.
the result curve between time to pass each 100 deals by machine on backtest shows polynomial curve and proofs that there are non-linearity because of back-test internal INCREASING process through time!
I need to know if I am right by this test or not?!
thanks everyone,
I run some tests with an EA to find out how run time for EA changes for equal count of deals.
EA runs on M1 time frame and there are thousand of deals for less than 3 months. I think number of data is good enough to consider bias as non-effective.
the result curve between time to pass each 100 deals by machine on backtest shows polynomial curve and proofs that there are non-linearity because of back-test internal INCREASING process through time!
I need to know if I am right by this test or not?!
thanks everyone,
This could be a memory problem, I mean is that when an optimization pass is run, it is only allocated a certain amount of memory and the longer it runs, that memory gets full which reduces the backtest speed
This could be a memory problem, I mean is that when an optimization pass is run, it is only allocated a certain amount of memory and the longer it runs, that memory gets full which reduces the backtest speed
yes, i have same guess.
but it will be a big issue when we are talking about a 10K pass for a simple optimization. last passes take tremendous time!!
It can be more easy if we have option. for very first assessments we do not need full menu of results. we just want to know if it worth to work more on that EA or not?!
If I extend testing time from 3 months to 6 months its run time extends 160% more!!! this is not good at all.
- Free trading apps
- Over 8,000 signals for copying
- Economic news for exploring financial markets
You agree to website policy and terms of use
As I see(please correct me if I am wrong), when a back test starts speed is very high and passing time and increasing positions/orders and all operations which may continuously are recoding on arrays and ... speed reduces dramatically!
Is it a good idea to give access to user to limit statistics and other results to speed up backtests?
this can be done even by categorizing backtest to 3 different levels like:
1- very simple backtest: not included anything just single item like profit or PF (what ever user selects)
2- minimum results like DD and Estimated Payoff and...
3- Perfect mode including all parameters available