Very nice Backtest

 
I'm consider a newbie in forex. Just wanted to ask. How much should i trust a backtest? Like let's say if a particular bot shown a very nice backtest for 10years and compare to maybe just 1year.

Which is more reliable? I was also taught that EA has expiry. Is this true?

Have also encountered before that life run didn't turn out to be as good as backtest. Or is backtest all just a scam?? Or did i just screwed up with the settings?
 

Hello ,welcome to Mql

I would prefer (personally) to see a specific time periods optimization perform on the next time period without optimization ,to get a grasp on how reliable the test is and how much can it help.

All that after ive made certain my backtesting criteria and conditions are exactly similar to live criteria and conditions.

consult this image :


 
Lorentzos Roussos:

Hello ,welcome to Mql

I would prefer (personally) to see a specific time periods optimization perform on the next time period without optimization ,to get a grasp on how reliable the test is and how much can it help.

All that after ive made certain my backtesting criteria and conditions are exactly similar to live criteria and conditions.

consult this image :

Thanks! Will take time to read them all up!

Have also attached a few image of my backtest results which i feel that may be too good to be true..

I've run it on life account but opened only 1 trade today.. base on my calculation it was supposed to open more trades on average daily basis.. or is it due to slow monday trading period? Also how much trust should i put into my backtest results?

 
Enrique Dangeroux:

It is not smart to post picture with your name and account
number.


Ahh.. removed it.. thanks..
So how much should i trust a backtest? I've recently just rent a EA showing good results. But author has been quite inactive so couldn't contact him.

 

Seeing a 90% modeling quality ,i would refrain from trading the system yet .

I think ,for starters ,moving the test to MT5 would be a good step .

My opinion -which is untested and not applied yet so you should not take it to heart- :

Technical :

  • Use real ticks to test
  • If you cant use real ticks then bring live execution to resemble the backtester
  • Make sure live execution conditions are simulated in the test
  • Use higher spreads than usual in the test (although i think you cant do that in mt5 unfortunately)
  • Make sure you are using data that would be available to the algorithm in real time 
edit : if you dont have the code some of the above is not feasible.
 
So 90% modeling quality is still not considered good enough? Cause i find it a hassle to download 99% modeling quality.. file size too big and somehow i got error after downloading one.. so i sticked to 90% modeling quality backtest.
 
Wei Zhen Kuan:
So 90% modeling quality is still not considered good enough? Cause i find it a hassle to download 99% modeling quality.. file size too big and somehow i got error after downloading one.. so i sticked to 90% modeling quality backtest.

It is difficult and not user friendly to achieve that 99.9% indeed . 

Has the author of the Ea any contact links ?,you can communicate with them there 

 
Tried pming the author.. no response. Other users also encountered the same issue. They mentioned that the EA isn't working. 

What are the factors as well for EA that's not working??
Broker?
Server?
Any other factors?
 
Wei Zhen Kuan:
Tried pming the author.. no response. Other users also encountered the same issue. They mentioned that the EA isn't working. 

What are the factors as well for EA that's not working??
Broker?
Server?
Any other factors?

That is a nuanced situation , anything can be an issue ,including the ea itself or the users configuration .
A moderator may advice properly on what to do in such a case as they have seen more similar cases . i assume