MT4 strategy tester is not reliable

New comment
 

hello

i'm doing a test on my EA on history data, let's say from 01/01/2010 - 01/01/2012 on EURUSD pair and i get some results

then without changing dates or anything in the code i run the same test again and i get totaly different results.

any solution to this problem or is there other reliable tool to do the test?

thanks in advance

 

BTW before asking i've updated my history data and i'm getting Ticks modelled of 90% and i'm using each tick testing every time

 
Check your spread.
 
clerin6:
Check your spread.

what do you mean by check my spread? it's on DEMO account...
 
redharel:

what do you mean by check my spread? it's on DEMO account...

If the Spread changes from Run 1 to Run 2 your results will be different.

The Spread at the start of the run is used for the whole run . . .

 
RaptorUK:

If the Spread changes from Run 1 to Run 2 your results will be different.

The Spread at the start of the run is used for the whole run . . .

Slippage is 0 in my EA so how spread is changing?


where am i suppose to define that? i can't find it on MT4?

 
redharel:

where am i suppose to define that? i can't find it on MT4?

Define what ? the Spread is the Spread . . you don't define it it's set by your Broker.

 
redharel:

Slippage is 0 in my EA so how spread is changing?


where am i suppose to define that? i can't find it on MT4?

What does Slippage have to do with Spread ?
 
If you want a fixed Spread while testing disconnect from your Broker . . . then the Spread won't change. Then you can run sever runs and see that the result doesn't change and then you can come back here and say sorry to the Strategy Tester. ;-)
 

The way I disconnect from my Broker . . .

File > Login, enter a false ID and click Login, it will fail and will disconnect you from your Broker.

 
RaptorUK:

The way I disconnect from my Broker . . .

File > Login, enter a false ID and click Login, it will fail and will disconnect you from your Broker.


before i'm trying that, it think it's not correct that my history back test is dependent on live spread, it suppose to be fix spread because we talking about history data.

that is my opinion at least

i'm checking it now

12
New comment
Reason: