Optimizing Indicators ?

New comment
 
How can I optimize Indicators that aren't in or part of any EA? The setup is for a manual trading system with up to 8 indicators attached to charts. I have the source code for some of them that are in the public domain, but not the primary ones which I got with the manual trading system that I bought. If it turns out to be profitable, I'd like to make it into an EA. But that is a way down the road and tentative depending on the results I get. Prior to that I want to optimize the external variables of the Indicators for specific, different pairs.
 

A system that uses 8 indicators is probably not going to work.

Explanation of above conclusion. Each indicator at any point in time either says trade or dont trade. which means 2 to the power 8 conditions of which (depending on the system) only 1 out of 256 options says it is time to trade as all have to be in agreement that it is time to trade. This is not the whole story as there are many ways to put systems together like some indicators to start trading and others used to add more trades when in a trading area.

Getting a 'best fit' for each individual indicator would not necessarily create a 'best fit' for the indicators working together in a system.

Suggest you look at http://forex-strategies-revealed.com/basic and do some research

 

I know that Kohonen maps or nearest neighbour alghorithms can be used for clustering(optimizing) but dont' ask me how.

 

Put the indicators in a DLL using Free Pascal where they go like a rocket and it is not that difficult to do. 7 bit has posted an example here

http://www.forexfactory.com/showthread.php?t=219576

I have taken this example and put my own indicator in it instead. It really is a doddle.

Once the indicator is working it can give the trade signals back to an EA so no need to mess about in MQ4 much just use it to open and close trade signals from the DLL indicator.

 
@Ruptor - Many thanks for this reference.
 

Howdy all,

The manual system has 3 indicators. I've been playing with adding the other's in various combinations to help get a handle and an idea on how the 'manual system' with its custom indicators works.

But I take your point and understand about to many indicators falling all over each other.

However the main point of this and what I am really trying to figure out is how to optimize indicators per say, regardless of the number of them, that are NOT in any EA. Which of coarse is what the strategy tester is made to work with.

So again: how does one go about 'optimizing' and back testing indicators when the strategy tester won't load and run them?

New comment