unstable testing results

New comment
 

I've been testing with the same parameters

no changes in the codes but the results are so different every time i run the strategy tester. I tested it with every tick mode but the profit is different every time i run

what is the problem here? 

 
doshur:

I've been testing with the same parameters

no changes in the codes but the results are so different every time i run the strategy tester. I tested it with every tick mode but the profit is different every time i run

what is the problem here? 

and in open only or 1m O/H/L/C mode? Still different everytime? every single thing else is the same? period/symbol/dates etc.? All in the same terminal with the same data right?
 
doshur:

I've been testing with the same parameters

no changes in the codes but the results are so different every time i run the strategy tester. I tested it with every tick mode but the profit is different every time i run

what is the problem here? 

Either one parameter changed (at least), or your code is non-deterministic
 

I did not change my code nor the testing period and I tested in every tick mode

I tested in the morning, I'm able to get $5xx but at night when I tested again, it is $4xx

The difference is so huge.

The only thing I changed is I switch to test USDJPY then switch back to EURUSD 

 
doshur:

I did not change my code nor the testing period and I tested in every tick mode

I tested in the morning, I'm able to get $5xx but at night when I tested again, it is $4xx

The difference is so huge.

The only thing I changed is I switch to test USDJPY then switch back to EURUSD 

did u get your settings loaded in from the optimizer by chance? I had an issue once where double clicking on the optimization results wasnt loading the values properly so I thought I was getting wrong results but actually it was just the settings werent loading in. Manually check the settings loaded properly? At first you said, its different every time. now you say it was different in the morning and didfferent at night. what happens when you run 2x in a row... both of the same now? Didnt sign in/sign up to another account that mightve changed the data? 
 
btw what broker?
 

Is it recommended to back test with my broker history data or MQ history data?

Should I delete the history of my broker and test again?

I also notice when I download EURCAD data, I ever saw a message like "synchronizing USDCAD" 

 
bendex77:
did u get your settings loaded in from the optimizer by chance? I had an issue once where double clicking on the optimization results wasnt loading the values properly so I thought I was getting wrong results but actually it was just the settings werent loading in. Manually check the settings loaded properly? At first you said, its different every time. now you say it was different in the morning and didfferent at night. what happens when you run 2x in a row... both of the same now? Didnt sign in/sign up to another account that mightve changed the data? 

In the morning I ran a couple of times and confirmed that I have $5xx and the code I have save a copy in notepad. The settings are default.

In the night I ran also a couple of times and have $4xx the same.

This had never happen before that's why I'm getting so confused.

maybe some more details on my code. I'm using M5 DEMA and H1 Envelopes. I coded to close buy when DEMA[1] is above Envelope upper[1] and DEMA[0] is below Envelope[0].

I'm running the tester on H1 timeframe. 

 
doshur:

Is it recommended to back test with my broker history data or MQ history data?

Should I delete the history of my broker and test again?

I also notice when I download EURCAD data, I ever saw a message like "synchronizing USDCAD" 

MQ goes back really far but only the majors are real data. EURCAD is calculated from USDCAD and EURUSD. I think "run once" tests are going to be the only thing that runs on your current broker data while I think optimizing needs to be MQ data to make it consistent across the cloud.
 
bendex77:
MQ goes back really far but only the majors are real data. EURCAD is calculated from USDCAD and EURUSD. I think "run once" tests are going to be the only thing that runs on your current broker data while I think optimizing needs to be MQ data to make it consistent across the cloud.

I see on the EURCAD...

My broker Alp*** opens one hour earlier than MQ data so if I optimize on MQ data and use on my broker then it is not the same. 

 

I forgot to mention I'm using MQ MT5 and key in my broker server ip manually to connect

is that going to be ok for optimization or should I download my broker MT5? 

12
New comment
Reason: