Backtesting/Optimization - page 69

New comment
 

Unfortunately you will not be able to do this in MT4 - at least not directly.

 

Problem with strategy tester

I have some EAs that will not test in MT4. You can see the progress bar going through the test, but when it's done, there is no graph or report, and the results are all zeros.

Any idea why this would affect some EAs, but not others?

Any assistance would be appreciated.

Rob

 
ChicagoRob:
I have some EAs that will not test in MT4. You can see the progress bar going through the test, but when it's done, there is no graph or report, and the results are all zeros.

Any idea why this would affect some EAs, but not others?

Any assistance would be appreciated.

Rob

The only reason I can think about is that there are no valid entry conditions for the EA to trade, your lot size is too big (try increasing the deposit on test or lowering the lot size). Another thing that you should check is if your stops are not too close.

Check the log files - if something is wrong, you'll find it there for sure.

 

TimeCurrent() in strategy tester

Is there any way to pull the current time in the tester instead of the last server time? When I run my EA, TimeCurrent() returns the last server time from when I logged on. I need it, or something else, to pull the current backtest time so I can backtest my EA.

Anyone got any ideas?

 

Pardo and tester

Hi people,

this is my first post here, but, as a reader, I'm quite an "addicted" of this forum, that I appreciate very much. It seems to be the place of the brightest mql developers, and I owe a lot of lessons to NewDigital, Igorad, Mladen and others, just to cite few.

The only little criticism is that sometimes things are a bit "cryptic", although this is mainly fault of my ignorance and not your mistake.

I am very interested in optimization topics, because I fear that most of the wonderful EA's around miss the objective just because there is nobody having the needed patience and pedantry to go through a full and somehow "scientific" development process (development comes AFTER invention).

Therefore I want just to ask (now) two things:

  1. Does anybody know R. Pardo's "The evaluation and optimization of trading strategies" (2008) book? What do you think about these ideas, methods and results?
  2. I think that the MT4-integrated tester is very powerful, but has some limitations for implementing the kind of strategies Pardo suggests. Just two issues. First: would it be possible "convincing" the tester to output backtesting results ordered by date of execution instead of number of order? (You know: it is different a system generating one trade per day, every day, than one generating a cluster of ten orders once, and then stay out of the market for two weeks before the next cluster... ). Second issue: is it possible to operate the backtester from some external script or even from a mql script? This could be e.g. interesting if you want to poerform genetical algorithm instead of grid optimization, or for step-forward analysis.

Nice to have comments! I apologize in advance if those questions were already discussed somewhere else, but with a so huge "mine of knowledge", sometimes finding information can be hard. Newdigital will help me surely with some link, eventually!

Bye

F

 

This is very nice!!!

 

Testing Renko?

I wonder if it's possible to use Strategy Tester on an offline chart. I have a Renko chart (named GBPUSD,m2) and am toying with some trading strategies. Does anyone know of a way to test an EA on an offline chart with a non-standard timeframe?

TIA

 
Lou G:
I wonder if it's possible to use Strategy Tester on an offline chart. I have a Renko chart (named GBPUSD,m2) and am toying with some trading strategies. Does anyone know of a way to test an EA on an offline chart with a non-standard timeframe? TIA

I don't think there is a way to do this in MT4, but I could be wrong.

The back tester doesn't really provide reliable results. I recommend the long, but exponentially more accurate forward test. This is the best way to achieve reliable results.

 
wolfe:
I don't think there is a way to do this in MT4, but I could be wrong. The back tester doesn't really provide reliable results. I recommend the long, but exponentially more accurate forward test. This is the best way to achieve reliable results.

You are absolutely right - I'm not a big fan of backtesting either but sometimes have used Strategy Tester just to get "the lay of the land" as far as a new strategy idea.

Thanks for responding,

Lou

 

Yeah, the strategy tester is good for quickly pointing out if your code is operating properly, so it is good for that.

Good luck with your system.

1...626364656667686970717273747576...95
New comment
Reason: