Greetings,

I am trying to automate a backtest for the proper 'mix' of quantity per target for best results in the strategy Analyzer. Basics are:

1. 4 targets (PT1, PT2, PT3, PT4)
2. PT4 == runner so quantity is 1
3. QuantityPT1 + QuantityPT2 + QuantityPT3 <= 9

So in the optimizer I have a variable (BestQuantity) I can set to between 3 >> 9, that would interact with the values selected in the optimizer for the QPT1, QPT2, QPT3 ... ie:

BestQuantity 5;9;1
Quantity ProfitTarget1 0;9;1
Quantity ProfitTarget2 0;9;1
Quantity ProfitTarget3 0;9;1
Quantity ProfitTarget4 1;1;1

So ... if at a point the optimizer has BestQuantity =9 and also selected 7 for QPT1, then it has to select QPT2 + QPT3 == total of 2 in whatever 'mix' and so on to allow the optimizer to select the best result for whatever the # of times the ProfitTargets happen to be filled.

I can setup properties, but when I tried to make " BestQuantity = (QPT1 + QPT2 + QPT3);" I have error on compile::
Only assignment, call, increment, decrement, and new object expressions can be used as a statement,CS0201

An assist would be greatly appreciated,
Jon

update:
I managed to have a successful compile, but the logic is flawed

This is from the Strategy Analyzer Settings:
BestQuantity 16 (3;9;1),
Quantity ProfitTarget 1 4 (3;7;1),
Quantity ProfitTarget 2 6 (0;6;1),
Quantity ProfitTarget 3 6 (0;6;1),
Quantity ProfitTarget 4 1

As you can see, the 'BestQuantity' optimized result is the addition of the maximum values selected for optimization (6+6+4 = 16), instead of the 9 intended.
Code used was in 'Initialize' :: BestQuantity = QPT1 + QPT2 + QPT3 .. BestQuantity and the QPTs are int variables

I'm having a hard time understanding your post. Perhaps screen shots, or c# code itself is in order.

But, I can tell you that when I design a strategy with multiple targets, I make those additional targets a cumulative total of the targets before them. This is just to make the optimizer make a bit more sense.

Target 1
Target 2 = Target1 + Target2
Target 3 = Target1 + Target2 + Target3

in code its basically

 

Code

t1 = T1ticks t2 = T1ticks + T2ticks t3 = T1ticks + T2ticks + T3ticks

This just makes the optimizer always go in a logical order, like target 3 is not smaller than target 1.

Mike

Big Mike 

I'm having a hard time understanding your post. Perhaps screen shots, or c# code itself is in order. But, I can tell you that when I design a strategy with multiple targets, I make those additional targets a cumulative total of the targets before them. This is just to make the optimizer make a bit more sense. Target 1 Target 2 = Target1 + Target2 Target 3 = Target1 + Target2 + Target3 in code its basically   Code t1 = T1ticks t2 = T1ticks + T2ticks t3 = T1ticks + T2ticks + T3ticks  This just makes the optimizer always go in a logical order, like target 3 is not smaller than target 1. Mike

Thanks for the input. I am trying to analyse for the best quantity of each target size. The actual size of each target isnt a factor AFAIK in the solution I am looking for. Target sizes in the strategy do progress in size.

Attached a screenshot of the optimizer setup and output showing the optimized # of trades per target add up to (here it is 7+5+1=13) which is more than the limitation set up by the 'QuantitySQ3' of 9, or the Strategy Analyzer optimized amount of 8 . On screen also are the only parts of the code that are actually involved in the 'QuantitySQ3' == the int variables for the quantity for each target and the total quantity allowed by QuantitySQ3.

Input is greatly appreciated!
Jon

Jon

NT support suggests that this is too complex for the backtester to complete, and doesnt even suggest it is possible with custom programming. I still have a feeling that a custom optimization type might accomplish the job, so I havent given up ... just have other more urgent code to build ...

IF anyone has thoughts ... someone that sees how useful this might be .. maybe we can build something

Jon

Big Mike 

But, I can tell you that when I design a strategy with multiple targets, I make those additional targets a cumulative total of the targets before them. This is just to make the optimizer make a bit more sense. This just makes the optimizer always go in a logical order, like target 3 is not smaller than target 1. Mike

Mike,

I always did wonder why you did that progression. For clarification, does this actually make a difference during optimization even if my targets would be in a pregressive order anyway? ... and if yes, in what way.

I am wonder though that 'maybe' the idea of adding quantity values together in some type of subroutine might be a step towards accomplishing the optimization of quantity per profit target I am looking for.

Regards,
Jon