Strange effect with asset() function

Hi,
When optimizing my script i faced the following weird issue:
The system shows different testing results depending on the way the currency pair is set. If it is set in concole form - one result, if it is set by calling asset() in script code - another result. As if in this case the price data is taken from some other source..(see screenshots attached). In both cases first i perform training and then testing. Could you explain what i do wrong?
Thanks in advance.

It might be usefull if you could post a minimal example code that can reproduce this behaviour so anyone can test and verify. Without code its really hard.

Hi, here is the code, which is simply the copy of workshop 7 with slight changes:

#include <profile.c>

function run()
{

MaxBars = 4320; //half a year
BarPeriod = 60; // H1
Weekend = 1; // don't merge Friday and Sunday bars
LookBack = 3; // only 3 bars needed
NumWFOCycles = 13;
Capital = 1000;
asset("USD/JPY");

set(RULES+FACTORS+TESTNOW);

if(Train) Hedge = 2; // for training, allow long + short

if(ReTrain) {
UpdateDays = -1;
SelectWFO = -1;
reset(FACTORS); // don't re-train factors
}
//
Margin = 0.5 * OptimalF * Capital * sqrt(1 + ProfitClosed/Capital);


if(adviseLong(PATTERN+2,0,
priceHigh(2),priceLow(2),priceClose(2),
priceHigh(1),priceLow(1),priceClose(1),
priceHigh(1),priceLow(1),priceClose(1),
priceHigh(0),priceLow(0),priceClose(0)) > 50)
reverseLong(1);

if(adviseShort() > 40)
reverseShort(1);

}

If you comment the line with asset(...) and set the pair in list box of console you`ll receive different result.

The results will be most likely the same, but the training files have different names dependent on whether it's a portfolio system with "asset" calls, or not. So make sure that you're using the right files.

Hi,
I encountered another issue with asset(). The following code does not cause error messages in train mode until asset() is called (if it is commented), and a pair is selected in the listbox. When called the message "Error 030 - Check lookback, settings, asset order" appear.

#include <profile.c>

function run()
{
StartDate = 2016;
EndDate = 2017;
BarPeriod = 60; // 1 day
LookBack = 0; //
NumWFOCycles = 10;

asset("AUD/USD");

set(RULES+PARAMETERS+TESTNOW);

if(Train) Hedge = 2;

Stop = 3*ATR(5);
Trail = optimize(13,8,55);
TrailLock = 1;
TrailSlope = 50;



LookBack = 3; // only 3 bars needed
if(adviseLong(PATTERN+2,0,
priceHigh(2),priceLow(2),priceClose(2),
priceHigh(1),priceLow(1),priceClose(1),
priceHigh(1),priceLow(1),priceClose(1),
priceHigh(0),priceLow(0),priceClose(0)) > 40)
reverseLong(1);

if(adviseShort() > 40)
reverseShort(1);

PlotWidth = 600;
PlotHeight1 = 300;
}

Here i set LookBack to 0 at the beginning to avoid problrms with calling ATR() and then change to 3 to set the number of candels for advise algorithm as mentioned in Workshop 7. Why that cease to work because of calling asse()?
Thanks in advance

Just as the error message says: You must set the lookback period before you can load the asset.

Thanks! Got it finally)

Hi, I hurried a bit to admit the issue solved. The problem is i can`t use in the same script ATR and adviseLong/Short because they need different LookBack values. In my case adviseLong/Short needs value 3 which is the number of bars in pattern, and ATR needs much more. First i change the value to suit ATR and after calling it i change the value to 3 for adviseLong/Short. But in this case Zorro returns error 030 (though LookBack-settings-asset sequense seems to be right).
Here is the code fragment (the whole text is similar to above posted):

LookBack = 500 (or 0);
NumWFOCycles = 10;

set(RULES+TESTNOW);
asset("EUR/USD");
Stop = 2*ATR(3);

LookBack = 3;
if(adviseLong(PATTERN+2,0,....

What is wrong here? How can i set a marketaware stop loss in case of using machine learning algo?

A variable can not be 3 and 500 at the same time, so use the higher value.

And then how AdviseLongShort will know that it should use 3(in this case) bars pattern for analysis? Or is 3 const value?

It won't. It follows the principle: Better too much than too little.