Linear regresion line backtesting

[argentarius]

Hello experts!
I want to play, experiment with linear regresion line but I ecountered the problem that I can not solve. It is about backtesting linear regresion line. I understand that I should use loop in my code so that the backest will be conducted bar by bar but it does seem to work at least not for me. If you guys guide me which direction I should proceed I would be very greatfull. Although my coding skills are very basic I fully understand array coding and looping, so any help from you will be appreciated.

[Cougar]

I want to play, experiment with linear regresion line but I ecountered the problem that I can not solve. It is about backtesting linear regresion line.

LinearReg() is just another array! Treat that the way you would backtest any other array.

How is LinRegSlope() calculated?

then you will see that the moving LinearRegression is an END POINT of estimate of slope and intercept at period bars. And that it changes on every bar. You will clearly see that only END POINT is on the LR line. Once you understand you will know why your formula is incorrect.

[argentarius]

Cougar!!!
This is my code, I have tried to make it as simple as possible

SetBarsRequired( sbrAll, sbrAll );            
MarginDeposit=3300;      
      
SetOption("MaxOpenPositions", 1 );       
       
SetPositionSize(1,spsShares);    
     
RoundLotSize=1;       
MarginDeposit=3200;       
PointValue=20;    
 SetTradeDelays(1,1,1,1);       
 BuyPrice=Open;  
 SellPrice=Open;  
 ShortPrice=Open;  
 CoverPrice=Open;
 ////////////////////////////////////////////////////////
_SECTION_BEGIN("Linear Regresion Line");   
 x = Cum(1);
lastx = LastValue( x ); Daysback = 15; aa = LastValue( LinRegIntercept( Close, Daysback) );
bb = LastValue( LinRegSlope( Close, Daysback ) );

y = Aa + bb * ( x - (Lastx - DaysBack) ); Plot( Close, "Close", colorBlack, styleCandle );
Plot( IIf( x >= (lastx - Daysback), y, -1e10 ), "LinReg", colorRed );
_SECTION_END();

//////////////////////////////////////////////////////////    
 TimeFrameSet(inDaily);
  
  _linregline = aa ; //I assume this is an END POINT of linregline        
TimeFrameRestore(); 
               
        
stsk=StochK(14,3);         
stsd=StochD(14,3,3);      
      
value1=70;         
value2=30;           
cond5=Ref(stsd,-1)<=value2  AND Ref(stsk<stsd,-1) AND  stsk>stsd;          
cond12 =Ref(stsd,-1)>=value1  AND Ref(stsd<stsk,-1) AND  stsd>stsk;           

 golong=C <TimeFrameExpand( _linregline,inDaily);  
 goshort=C >TimeFrameExpand( _linregline,inDaily);        
 
 longtrade= cond5 AND golong; 
 shorttrade= cond12 AND goshort; 
  ///////////////////////////////////////////////////  
 
 Buy= longtrade;   
 Sell= shorttrade ;  
 short=shorttrade; 
 Cover = longtrade;   
 ///////////////////////////////////////
 Short=ExRem(Short,Cover);     
 Cover=ExRem(Cover,Short);      
 Buy=ExRem(Buy,Sell);        
 Sell=ExRem(Sell,Buy);      

I treated it as another array, but I still do not have desired results. Once again. I want the price to be below daily linear regresion line on the particular date (daily bar), when I want to go long and oposite, the price should be above Daily linear regresion line when I want to go short. but it was not always the case. I checked it with the bar reply. I went back to the past trades with the bar reply and I discovered the problem. My question is where is my mistake? Is there problem in my code, or rather in the method I use ?

[Cougar]

Is there problem in my code

Simply use LinearReg() - compress it to Higher timeframe and then expand. Moreover, that code is not generating signals. Quoting Tomasz:

Keep signal active for more than one bar

State_Form = Close > MA( Close, 10 ); // state form keeps 'true' for many bars
Signal_Form = Cross( Close, MA( Close, 10 ) ); // signal (pulse) form - one bar only

And while expanding an array, please understand the difference between expandFirst and expandLast.

[argentarius]

Thank you Cougar for your response. I understand now,.I can backtest linreg line as any other array becouse it is a curve in backtesting.

[Cougar]

@argentarius,

In the opening post of this thread and the code that you shared, you've clearly depicted:

It is about backtesting linear regresion line.

Quoting from LinearReg( ARRAY, periods ) definition:

Linear regression line end-point value is calculated according to a + b * x (where a and b are intercept and slope of linear regression line) from the ARRAY using periods range.

It is ambiguous (thus incorrect) to state:

becouse "it" is a curve in backtesting.

LinearReg() in our topic of discussion is not curve-fitting, because it is end point (not forward looking) and has no predictive power. Once again exhibiting from the same link LinearReg( ARRAY, periods ):

So, YES you can treat this LinearReg() like you would treat any other array while backtesting.