Tester optimization error

New comment
 

Hi, 

i want to test my expert advisor with optimization over all the market symbols. 

But when i start the test, i get the error "optimization cannot be startet". Thats all....


Am i missing something in my expert advisor? What is needed to run the optimization?


Kind regards, 

Robert

 

Ok, i added an input variable and it finally works. But i dont know why this changes anything, because the variable doesnt do anything :D


But maybe this thread can help someone with the same problem in the future, cause i googled and found nothing :)


Greets 

Robert

 
Hey im having the same problem, can you give me the input variable code? Thanks
 

No inputs: there is nothing to optimize — no point in starting.

An unused input: there is nothing to optimized — Tester can't tell.

input int not_used;
 
I defined what William said and backtest started.. not sure how but a BIG THANK YOU !! 
input int not_used;
 
Hi - I'm new on here. I couldn't work out how to post a new forum question, any help on that would be appreciated.

My main question though is about CPU. I'm developing quantitative strategies, backtesting and optimising them for a partners and clients. As I'm trying to up the scale of my work, I'm looking to invest in new hardware for my computer. Biggest issue I havs is which CPU to get. I know most CPUs are developed for 3D rendering but I was wondering if there's a CPU which was designed for what I need, optimisation, to get best bang for buck. I'm looking to spend up to 1500 as its a business expense so if anyone has a list of their recommendations I'd appreciate it. Thanks.
 
Herbert321: 
Or even a place to check out!
 
The more cores you have the more optimization runs can be done simultaneously.
For the sake of writing code, any cpu will do the job. For shortening the compile times, the single thread speed is important.

Hope this helps.

Also, the faster your memory, the less latency in back testing and optimization you will have.

But in the end, a good coded EA will be the magic to get really fast results.

If you intend to do lots of optimization runs over a course of multiple days, you'd want ECC memory and a server class cpu. - For reliability.

In case you are using OpenCL, you will need a GPU or multiple of them. Here you can go and look for GPUs used for neural networks, as this is the closest comparable load to whàt you will probably be doing.








New comment
Reason: