Optimization gives random results?

 

Is it possible that tester yields different results on different runs ?

I mean I ran an optimization, non genetic over some variables and some time period about a week ago.

Now im running the same optimization, same variable ranges, same time preiod, same EA (everything is exactly the same) but the results are different even for identical combination of the input variables.

There is a strong correlation though between the results in the first and in the second run, but shouldn't the results be identical ?

Thanks very much for any hint on this...

EN.-

 
Everything is exactly the same ? how are you fixing the Spread so it is exactly the same as a week ago ?
 

Ups, didnt think on that. My inputs are all the same so i thought the results would be the same.

So i guess spread used in tester calculations is not historically accurate.

Can it change drastically the results of two runs separated by only one week ? How does this work ? Care to explain to me in two lines ? Thank you very much.

 
esteban:

Ups, didnt think on that. My inputs are all the same so i thought the results would be the same.

So i guess spread used in tester calculations is not historically accurate.

Can it change drastically the results of two runs separated by only one week ? How does this work ? Care to explain to me in two lines ? Thank you very much.

The spread used is the current spread for the pair you are testing on at the time you start the test and it remains fixed for the entire run.

Yes it can drastically change the results, it can change the results for 2 runs 5 mins apart . . . how much depends on your EA.

 

Ok, thanks very much for your help!

 
esteban:

Ok, thanks very much for your help!

I use this simple method to disconnect from my Broker and keep my spread constant. Goto File > Login . . . in the Longin: box I enter some random numbers, then click Login . . . the login fails and you are now disconnected from your Broker.
 

Ok, I will try that. Thanks again. EN.-