MQL4 StrategyTester relevancy and different timeframe

 

Hello guys,

 I wonder, how relevant and how well are tests done by StrategyTester in MT4, whether I can rely on that tests (I test on ticks)

And I'd also like ask, whether I can test on Week timeframe on MT4 somehow, thanks!

 
PeterBocan: I wonder, how relevant and how well are tests done by StrategyTester in MT4, whether I can rely on that tests (I test on ticks)

That is a very vague and philosfical question! The answer is that it depends:

  • it depends on your strategy,
  • it deepens on how good your code is (is it well written or is it "buggy"),
  • it depends on how you interpret and apply the results.

Good results in a back-test is not a guarantee of future good results, because markets change, but back-tests are essential in order to test functionality as well as feasibility of the strategy. You should always conduct in-sample and out-of-sample tests as well as forward tests. In essence, back-tests are only as good as the knowledge and experience of the author or tester. In the hands of someone clueless on the matter it would be the same as no tests at all, if not worse.

So, the question should be, how relevant is your knowledge and experience in carrying out back-tests as well as analysing and understanding the results!


PeterBocan: And I'd also like ask, whether I can test on Week timeframe on MT4 somehow

By default, the Strategy Tester no longer has "Weekly" back-tests, but with the aid of 3rd party tools, you can still accomplish this.

Since you mentioned using Tick Data, I will assume you are using TickStory (or something similar), which can produce Weekly FXT files. So, just rename the "Weekly" FXT file as the "Daily" FXT file and run it as "Daily". The result will be "Weekly" back-testing.

 
  • it depends on your strategy,


My strategy is basically set and wait (trend following), that's why I try to test on daily/weekly.

  • it deepens on how good your code is,

My code is rather simple, it is based on statistics and pre computation of probability of some events, so a lot of work is done even before the algorithm. 

  • it depends on how you interpret and apply the results.
I don't really understand what do you mean by that...
 
PeterBocan: I don't really understand what do you mean by that...

Well, then there is your answer. At the moment a back-test will be useless to you, if you don't understand the results.

It is important that you understand the results in order to be able to detect:

  • to detect if your strategy is being properly applied or not,
  • to detect if the code is functioning correctly,
  • to detect if it can adapt to different market conditions, such as changes in spread, slippage, etc.
  • to see if it is profitable, in accordance with the the various metrics available to you:
    • Profit
    • Profit Factor,
    • Draw-down,
    • Recovery,
    • Risk:Reward ratio,
    • MAE/MFE
    • etc.
 

 Any reliable source for backtesting

 
PeterBocan:

 Any reliable source for backtesting? 

I don't know of any single centralised documentation source that covers all that I have mentioned. There is information available on the web but it is scattered over many sites and sources.

In my case, I acquired the knowledge by experimentation and doing the research on my own.

 
FMIC:

I don't know of any single centralised documentation source that covers all that I have mentioned. There is information available on the web but it is scattered over many sites and sources.

In my case, I acquired the knowledge by experimentation and doing the research on my own.

It would be great to have one, okay, I'll get into googling.