Here's the EasyLanguage code for John Ehlers' 4 Pole Gaussian Filter:

Four Poles: f = alpha^4g + 4(1-alpha)f[1] - 6(1-alpha)^2f[2] + 4(1-alpha)^3f[3] - (1-alpha)^4f[4]
g=price and where alpha = =-Beta+SQR(Beta^2 +2*Beta), and Where Beta = (1 – cos(symbol)) / (1.414^2/N – 1), where symbol = 2pi/P, N = # of poles in the filter, in this case 4, and P = Period, i.e 15

Here's what I came up with: (I just solved what 2pi b/c I don't know the function to use for pi, and I solved what 1.414^(2/N); N=4(4 Pole), so equals 1.18911732 minus 1 from the EasyLanguage formula above. The problem I'm experiencing is that I get a legit plot for 15 period, but once I adjust the plot, i.e move to 20 period, it goes insane, rendering it useless for higher than 20 period. Reason? Any suggestion would be awesome.

input period = 15;
input paintbars = {"Yes", default "No"};
def price = vwap;
def symbol = (6.28318531) / period;
def beta = (1 – Cos(symbol)) / (0.18911732);

def alpha = -beta + Sqr(power(beta,2) + (2 * beta));
rec Gaussian = (Power(alpha, 4)) * price + (4 * (1 - alpha)) * Gaussian[1] - 6 * Power(1 - alpha, 2) * Gaussian[2] + 4 * Power(1 - alpha, 3) * Gaussian[3] - Power(1 - alpha, 4) * Gaussian[4];

plot GF= Gaussian;

The thing to remember with EasyLanguage code is that it automatically converts degrees to radians whereas thinkscript does not. Here's a Gaussian 2-pole filter in thinkscript. Hope this helps.


declare upper;
input period = 9;

script DegToRad {

input degrees = 0.0;
plot DegToRad = degrees * Double.Pi / 180;

};

def beta1 = 2.415 * (1 - Cos(DegToRad(360.0 / period)));
def alpha = -beta1 + Sqrt(beta1 * beta1 + 2 * beta1);

def mCoef1 = alpha * alpha;
def mCoef2 = 2 * (1 - alpha);
def mCoef3 = -(1 - alpha) * (1 - alpha);

rec mFilterSeries = mCoef1 * hl2 + mCoef2 * mFilterSeries[1] + mCoef3 * mFilterSeries[2];

plot GaussianFilter = if IsNaN(close[2*period]) then Double.NaN else mFilterSeries;

GaussianFilter.SetDefaultColor(Color.yellow);