Please help! Strategy tester results vary dramatically when the tester's period is not the same as indicator

[John2010]

Hi guys,

I am testing a strategy on MT4, and have noticed a very strange phenomena.

When In the strategy tester I choose "H1 Period", and in the EA code I use data from 5 min period indicators, the results are steady and great, it grows 3K deposit into 7.5 millions! But when I use the same period in strategy tester and for the indicator, the results will be fluctuating and non-sense.


Note: I am using control points model. Every tick and open prices return fluctuating results too. I know every tick is more precise, but I am wondering how different the results could be! Also, dont forget the main question: Great results come when I choose 1 hour period in the tester and 1 minute in the indicator (inside the code), any other combination is non-sense.


First, please give me your thoughts on this, I am puzzled, what ma I missing?

What is the impact of chart period on the EA? Since I am giving a fixed period length to the indicator in the code.


Please look at the screen shots.

[William Roeder]

John2010:
I am testing a strategy on MT4, and have noticed a very strange phenomena.


Note: I am using control points model.

First, please give me your thoughts on this, I am puzzled, what ma I missing?

1. Why did you post your MT4 question in the MT5 General section instead of the MQL4 section, (bottom of the Root page?)
             General rules and best pratices of the Forum. - General - MQL5 programming forum?
   Next time post in the correct place. The moderators will likely move this thread there soon.

2. Stop using them. All you are doing is putting a band-Aid on the real problem; you've solved nothing.
   
3. Have you downloaded complete history M1, M5, M15, and H1 for your entire test period. Otherwise, you have garbage.
   

[John2010]

William Roeder:

1. Why did you post your MT4 question in the MT5 General section instead of the MQL4 section, (bottom of the Root page?)
             General rules and best pratices of the Forum. - General - MQL5 programming forum?
   Next time post in the correct place. The moderators will likely move this thread there soon.

2. Stop using them. All you are doing is putting a band-Aid on the real problem; you've solved nothing.
   
3. Have you downloaded complete history M1, M5, M15, and H1 for your entire test period. Otherwise, you have garbage.
   

Thanks dear William,

Sorry about posting in the wrong place. I thought it belongs to the General Category.

What do you mean by no.2 "stop using them ...", I dont understand. Would you please clarify?

regarding the history, yes I have downloaded the complete history (M1, M5, etc) from MT4 database, how reliable is Meta quotes data? would it be better to get data from Tickstory or other companies?

My question is this: In the EA I am using 1 min/1 min Bollinger bands, but in the tester there is a Period menu (above Speard). When I choose 1 hour period, the results would be stunning! when I choose same period as bollinger bands in the EA the result would be medicore. Also, it goes to zero when I test it on every tick model!

I really appreciate your help!

[amando]

John2010:

Thanks dear William,

Sorry about posting in the wrong place. I thought it belongs to the General Category.

What do you mean by no.2 "stop using them ...", I dont understand. Would you please clarify?

regarding the history, yes I have downloaded the complete history (M1, M5, etc) from MT4 database, how reliable is Meta quotes data? would it be better to get data from Tickstory or other companies?

My question is this: In the EA I am using 1 min/1 min Bollinger bands, but in the tester there is a Period menu (above Speard). When I choose 1 hour period, the results would be stunning! when I choose same period as bollinger bands in the EA the result would be medicore. Also, it goes to zero when I test it on every tick model!

I really appreciate your help!

You cannot do multi timeframe backtest on mt4

[William Roeder]

sure you can. Just not weekly, or Monthly.
          'Testing Features and Limits in MetaTrader 4' - MQL4 Articles 2012(The comment about "Zero bar of another timeframe," is no longer valid.)

[John2010]

William Roeder:
sure you can. Just not weekly, or Monthly.
          'Testing Features and Limits in MetaTrader 4' - MQL4 Articles 2012(The comment about "Zero bar of another timeframe," is no longer valid.)

Hi, I wish you could give me a clearer answer.

In the strategy tester above the spread menu, I choose 1Hr, but in the EA, I am using indicators with 5 minute intervals. This would lead to a great stable results. But if I use 5 minutes as both it would lead to a mediocre results. Could you explain what's happening?