troubles DataSeries

[Nikoli]

Hi

I want modify DataSeries and get forecast with LinReg
I wrote next code

protectedoverridevoid Initialize(){

CalculateOnBarClose = true;
Overlay = false;
PriceTypeSupported = false;

myDS = new DataSeries(this);}

protectedoverridevoid OnBarUpdate(){
int m = 5;
if (CurrentBar < m) return;

if (CurrentBar % m == 0){
for (int i=0; i<m; i++) {
myDS.Set( i, m+i);
Print("i= "+i+" m+i="+(m+i)+" myDS.Get(myDS.Count-i)= "+myDS.Get(myDS.Count-i));
}
double value = LinReg(myDS, m-1)[0];
Print(" The current LinReg "+m+" value is " + value.ToString());
}

and after a few bars got printed

i= 0 m+i=5 myDS.Get(myDS.Count-i)= 0
i= 1 m+i=6 myDS.Get(myDS.Count-i)= 5
i= 2 m+i=7 myDS.Get(myDS.Count-i)= 6
i= 3 m+i=8 myDS.Get(myDS.Count-i)= 7
i= 4 m+i=9 myDS.Get(myDS.Count-i)= 8

The current LinReg 5 value is -37

Why if i= 1 and m+i=6 the myDS.Get(myDS.Count-i)= 5 if code

myDS.Set( i, m+i);

and
how I can get forecast for only the last members myDS
example 5 from myDS.Count- 4 to myDS.Count-0 ?

Thkx

[NinjaTrader_Austin]

Nikoli, from what I can tell, everthing is behaving as expected. You are using the .Set overload (DataSeries.Set(int barsAgo, double value)) that specifies how many bars back you want to set a value. In the for loop, the code just fills your DataSeries with data (up to i bars back) and then prints the information.

Keep in mind that DataSeries are reverse indexed, that is the most recent value is stored at the index value of 0 and the oldest value is stored at DS.Count - 1.

You may want to review the documentation for DataSeries.

[Nikoli]

Nikoli, from what I can tell, everthing is behaving as expected. You are using the .Set overload (DataSeries.Set(int barsAgo, double value)) that specifies how many bars back you want to set a value. In the for loop, the code just fills your DataSeries with data (up to i bars back) and then prints the information.

Keep in mind that DataSeries are reverse indexed, that is the most recent value is stored at the index value of 0 and the oldest value is stored at DS.Count - 1.

Austin, from the first I wrote

for (int i=0; i<m; i++) {
myDataSeries.Set( i, m+i);
Print("i= "+i+" m+i="+(m+i)+" myDataSeries.Get(i)= "+myDataSeries.Get(i)+" myDataSeries[i].ToString()= " + myDataSeries[i].ToString());}
double value = LinReg(myDataSeries, m-1)[0];
Print(" The current LinReg "+m+" value is " + value.ToString());

and got next

i= 0 m+i=5 myDataSeries.Get(i)= 0 myDataSeries[i].ToString()= 5
i= 1 m+i=6 myDataSeries.Get(i)= 9 myDataSeries[i].ToString()= 6
i= 2 m+i=7 myDataSeries.Get(i)= 4 myDataSeries[i].ToString()= 7
i= 3 m+i=8 myDataSeries.Get(i)= 3 myDataSeries[i].ToString()= 8
i= 4 m+i=9 myDataSeries.Get(i)= 2 myDataSeries[i].ToString()= 9
The current LinReg 5 value is 10243,2

where error?
example

myDataSeries.Set( 1, 6);
Print("i= "+1+" m+i="+6+" myDataSeries.Get(i)= "+myDataSeries.Get(1));

in penultimate member set 6
Why penultimate .Get(1) return 9 and [1] return 6 ?

and

how I can get forecast for only the last 5 members myDataSeries?

[NinjaTrader_Austin]

Nikoli, please only use the brackets to access DataSeries values, not the .Get methods. I believe that will help clear up confusion. In addition, unless absolute necessary, setting historical DataSeries values in a for loop isn't recommended. Wrong values can be set very easily.

Now I'm not exactly sure what you mean by forecasting only the last 5 members of myDataSeries, but if you wanted to run a linear regression on the most recent 5 values of your DataSeries you could do this:

double forecasted_value = LinReg(myDataSeries, 5)[0];

[Nikoli]

Nikoli, please only use the brackets to access DataSeries values, not the .Get methods. I believe that will help clear up confusion. In addition, unless absolute necessary, setting historical DataSeries values in a for loop isn't recommended. Wrong values can be set very easily.

Now I'm not exactly sure what you mean by forecasting only the last 5 members of myDataSeries, but if you wanted to run a linear regression on the most recent 5 values of your DataSeries you could do this:

double forecasted_value = LinReg(myDataSeries, 5)[0];

as a result we have very simple code

protectedoverridevoid OnBarUpdate(){
int m = 5;
if (CurrentBar < m) return;

if (CurrentBar % m == 0){
Print(Time[0].ToLongTimeString()+" ============");
for (int i=0; i<m; i++) { // input and output for check m=5 rececent value
myDataSeries.Set( i, m+i);
Print("i= "+i+" m+i="+(m+i)+" myDataSeries[i].ToString()= " + myDataSeries[i].ToString() );
}
// output forecast for check
double value = LinReg(myDataSeries, m)[0];
Print(" The current LinReg "+m+" value is " + value.ToString());
}

and print-results of the code

3:14:38 ============
i= 0 m+i=5 myDataSeries[i].ToString()= 5
i= 1 m+i=6 myDataSeries[i].ToString()= 6
i= 2 m+i=7 myDataSeries[i].ToString()= 7
i= 3 m+i=8 myDataSeries[i].ToString()= 8
i= 4 m+i=9 myDataSeries[i].ToString()= 9
The current LinReg 5 value is 5
3:14:48 ============
i= 0 m+i=5 myDataSeries[i].ToString()= 5
i= 1 m+i=6 myDataSeries[i].ToString()= 6
i= 2 m+i=7 myDataSeries[i].ToString()= 7
i= 3 m+i=8 myDataSeries[i].ToString()= 8
i= 4 m+i=9 myDataSeries[i].ToString()= 9
The current LinReg 5 value is -13
3:14:58 ============
i= 0 m+i=5 myDataSeries[i].ToString()= 5
i= 1 m+i=6 myDataSeries[i].ToString()= 6
i= 2 m+i=7 myDataSeries[i].ToString()= 7
i= 3 m+i=8 myDataSeries[i].ToString()= 8
i= 4 m+i=9 myDataSeries[i].ToString()= 9
The current LinReg 5 value is -31

We can see:
- 5 the most recent values is stable (9, 8, 7, 6, 5)
- but forecast on 5 the most recent values change on each step - why ?
- linear regression from (9, 8, 7, 6, 5) must be 4 - why we have 5 as better result ?

[NinjaTrader_Bertrand]

Nikoli, have you tried another method as such as the SMA on your custom dataseries for example as well? With which CalculateOnBarClose setting are you running this test?

[Nikoli]

Yes, I tried different metods, but results was bad.
Look, It is all code

protectedoverridevoid Initialize(){
CalculateOnBarClose = true;
Overlay = false;
PriceTypeSupported = false;
myDataSeries = new DataSeries(this);
}

protectedoverridevoid OnBarUpdate(){
int m = 5;
if (CurrentBar < m || (Historical)) return;

if (CurrentBar % m == 0){
Print(Time[0].ToLongTimeString()+" ============");
for (int i=0; i<m; i++) { // input and check m=5 rececent value
myDataSeries.Set( i, m+i);
Print("i= "+i+" m+i="+(m+i)+" myDataSeries[i].ToString()= " + myDataSeries[i].ToString() );
}
double value = LinReg(myDataSeries, m)[0];
Print(" The current LinReg "+m+" value is " + value.ToString());
Print(" The current LinReg "+m+" SMA(myDataSeries,1) value is " + LinReg(SMA(myDataSeries,1), m)[0].ToString());
}

We get next print-results

16:24:02 ============
i= 0 m+i=5 myDataSeries[i].ToString()= 5
i= 1 m+i=6 myDataSeries[i].ToString()= 6
i= 2 m+i=7 myDataSeries[i].ToString()= 7
i= 3 m+i=8 myDataSeries[i].ToString()= 8
i= 4 m+i=9 myDataSeries[i].ToString()= 9
The current LinReg 5 value is 5
The current LinReg 5 SMA(myDataSeries,1) value is 5
16:24:07 ============
i= 0 m+i=5 myDataSeries[i].ToString()= 5
i= 1 m+i=6 myDataSeries[i].ToString()= 6
i= 2 m+i=7 myDataSeries[i].ToString()= 7
i= 3 m+i=8 myDataSeries[i].ToString()= 8
i= 4 m+i=9 myDataSeries[i].ToString()= 9
The current LinReg 5 value is -13
The current LinReg 5 SMA(myDataSeries,1) value is -0,6
16:24:12 ============
i= 0 m+i=5 myDataSeries[i].ToString()= 5
i= 1 m+i=6 myDataSeries[i].ToString()= 6
i= 2 m+i=7 myDataSeries[i].ToString()= 7
i= 3 m+i=8 myDataSeries[i].ToString()= 8
i= 4 m+i=9 myDataSeries[i].ToString()= 9
The current LinReg 5 value is -31
The current LinReg 5 SMA(myDataSeries,1) value is -6,6

Pls, copy the code on Y comp and test indicator.

I think, CalculateOnBarClose not must influence on calculation.
And straight use myDataSeries must get the most correct results.

[NinjaTrader_Bertrand]

Thanks Nikoli, I'm getting the same results here on my end, checking into.

[Nikoli]

Thanks Nikoli, I'm getting the same results here on my end, checking into.

Thanks Bertrand, I'm waiting Y results.
Can I help Y ?

[NinjaTrader_Bertrand]

Nikoli, you're weclome - you will need to check into your loop logic for assigning the custom data series values, as for example my snippet below works as expected for me on ES in testing -

myDataSeries.Set(Bars.BarsSinceSession);
		
		double value0 = SMA(myDataSeries, 10)[0];
		double value1 = LinReg(myDataSeries, 10)[0];
		
		Print(" The current value is " + myDataSeries[0].ToString());
		Print(" The current SMA value is " + value0.ToString());
		Print(" The current LinReg value is " + value1.ToString());

Can you please confirm if this is also the case on your end?

Thanks

[Nikoli]

I pasted Y's code in my

protectedoverridevoid OnBarUpdate(){
int m = 10;
if (CurrentBar < m || (Historical)) return;

if (CurrentBar % m == 0){
Print(Time[0].ToLongTimeString()+" ============");
for (int i=0; i<m; i++) { myDataSeries.Set( i, m+i); }

double value = LinReg(myDataSeries, m)[0];
Print(" The current LinReg "+m+" value is " + value.ToString());
Print(" The current LinReg "+m+" SMA(myDataSeries,1) value is " + LinReg(SMA(myDataSeries,1), m)[0].ToString());

myDataSeries.Set(Bars.BarsSinceSession);

double value0 = SMA(myDataSeries, 10)[0];
double value1 = LinReg(myDataSeries, 10)[0];

Print(" The current value is " + myDataSeries[0].ToString());
Print(" The current SMA value is " + value0.ToString());
Print(" The current LinReg value is " + value1.ToString());
}

correct?

and got print

21:22:12 ============
The current LinReg 10 value is 10
The current LinReg 10 SMA(myDataSeries,1) value is 10
The current value is 13442
The current SMA value is 1357,7
The current LinReg value is 10
21:22:40 ============
The current LinReg 10 value is -4676,78181818182
The current LinReg 10 SMA(myDataSeries,1) value is -13432,3454545455
The current value is 13452
The current SMA value is 1345,2
The current LinReg value is -4676,78181818182
21:23:04 ============
The current LinReg 10 value is -9367,01818181818
The current LinReg 10 SMA(myDataSeries,1) value is -26885,3454545455
The current value is 13462
The current SMA value is 1332,7
The current LinReg value is -9367,01818181818

[NinjaTrader_Bertrand]

Nikoli, my point was simply that your custom loop filling the dataseries is the issue here, when you don't use this but for example the generic one I provided the values hit home as expected.

[Nikoli]

Nikoli, my point was simply that your custom loop filling the dataseries is the issue here, when you don't use this but for example the generic one I provided the values hit home as expected.

Bertrand, pls give me full code Y's indicator.

[NinjaTrader_Bertrand]

Sure, please try for example the attached on 6.5

[Nikoli]

Nikoli, my point was simply that your custom loop filling the dataseries is the issue here, when you don't use this but for example the generic one I provided the values hit home as expected.

Bertrand, task is:
1. modify several recent memberS ( NOT ONE THE LAST )
2. and forecast from them

Pls, write Your indicator with the conditions.

I think, not importance how we modify several recent members - in loop or in few lines of code. Because we checked it in Print() - values is correct.
We can use code with modify in few lines from Your example indicator with LinReg().
Note: modify several recent members is correct and legal.

But, what happen next when we try forecast with LinReg() ?

Where error ?