SMA and DataSeries

[Arnon_B]

Hi, Please consider the following problem I have:
private DataSeries ds;

void Initialize()
{
ds = new DataSeries(this); // Create DataSeries
}

void OnBarUpdate()
{
for(int i = 0; i <= 20; i++)
ds.Set(i,i); // fill the series again
double f = SMA(ds, 3)[0]; // SMA of last 3 bars
Print(f);

}

The problem is, that only at the first time f is correct (f = 1) i.e. (0 + 1 + 2) / 3
and on every new Print, f is not the SMA of those last 3 bars in ds, which are always 0, 1, 2....
also, I see that ds is growing itself with each new BarUpdate, why is that.
Thanks in advance.
Arnon.

[NinjaTrader_ChelseaB]

Hi Arnon_B,

I'm looking into this and I will have an answer for you shortly.

I appreciate your patience.

[NinjaTrader_ChelseaB]

Hi Arnon_B,

Thanks for your patience as I looked this over.

The issue here is with the SMA. The SMA does not expect you to change the historical values of the fed dataseries.

The line
double last = Value[1] * Math.Min(CurrentBar, Period);
is what is causing the issue.

Instead of recalculating the values for the entire series after every bar, the SMA uses the previous value of the SMA (Value[1]) * the period (so the weight is correct) and then adds the last value from your input dataseries (Input[0]) and divides by the period.

So basically, if you plan to constantly adjust your dataseries historically, you would need to recode the SMA so that it is forced to recalculate the entire historical series on every OnBarUpdate.

[Arnon_B]

Thanks a lot.
Arnon.

[koganam]

Hi Arnon_B,

Thanks for your patience as I looked this over.

The issue here is with the SMA. The SMA does not expect you to change the historical values of the fed dataseries.

The line
double last = Value[1] * Math.Min(CurrentBar, Period);
is what is causing the issue.

Instead of recalculating the values for the entire series after every bar, the SMA uses the previous value of the SMA (Value[1]) * the period (so the weight is correct) and then adds the last value from your input dataseries (Input[0]) and divides by the period.

So basically, if you plan to constantly adjust your dataseries historically, you would need to recode the SMA so that it is forced to recalculate the entire historical series on every OnBarUpdate.

Or you can Dispose() the SMA() each time, before you call it, so that it is always calculated afresh.

[Arnon_B]

Thanks a lot.
Arnon.