VWAP Trend Skew For ThinkOrSwim

[ysanchezmd]

Hi Mashume, can you help me to thinkscript the computation of the third central moment in the skewed distribution I am not educated in statistics, so I needed to read a few articles in Wikipedia. plus Jerry Perl article Trading with Market Statistics on Traders Laboratory. I had tried the formula (skew = vwap - vpoc /sd) but this may be better.


CM3 = sum(PROBi * (Pi - VWAP)^3),

where

i is going through all prices in range (i.e. all rows in Volume Distribution Function)

CM3 ... the 3rd Central Moment

PROBi = Vi / V ... ith price probability (Volume per ith price / Total Volume)

Pi ... ith price in the Vol. Dist. function



Then the Skew would be calculated as

Skew = CM3 / SD^3

I point of view is if we know the prediction of the price for the next hour and the skew, that will be a better edge trading the instrument. If the prediction is up and the skew is above vwap ( positive) then we has a bullish bias ....

Sorry my English is not very good looking ..........

As per Jerry Perl the distribution function itself tells you the trend bias and strength based on the strength of the skewThe VWAP/VPOC relationship determines the type of distribution (skewed/symmetric). Ignoring for the moment that as price moves around and volume trades that the VWAP and VPOC change and the distribution will probably change. Let's just say we were pretty sure that the type of distribution was not going to change for the next hour( Algorithmic Prediction). I think the distribution function itself tells you the trend bias and strength based on the strength of the skew, if Algorithmic Prediction is Up and the distribution had positive skew, you would go long every time the price action dropped below the VPOC and vice versa for negative skew. Similarly for no skew distribution, always trade toward the VWAP. This is the classical reversion to the mean theory.

 

[mashume]

@ysanchezmd
After reading your bit, I really like the idea of determining probability at a given price by the volume at that price vs total volume... very nice idea!!!

So I've come up with this bit of code:

declare lower;

input length = 20;

def V = sum(VOLUME, length);
def VWAP = VWap();

def CM3 = fold i = 0 to length with s do s + ((getValue(VOLUME, i) / V) * Power((GetValue(CLOSE, i) - VWAP), 3));

def SD = StandardDeviation(length = length);
plot skew = - CM3 / Power(SD, 3);
plot zero = 0;

The only thing I'm not sure about is whether to subtract VWAPi or VWAP in the definition of CM3. I think the indicator looks more useful subtracting VWAP, but I think it's mathematically more correct to subtract VWAPi. If you want to see that in action, use this line instead:

def CM3 = fold i = 0 to length with s do s + ((getValue(VOLUME, i) / V) * Power((GetValue(CLOSE, i) - GetValue(VWAP, i)), 3));

But, as I said, I think the one in the first code is more useful looking.

NOTE: I inverted the skew plot as it makes more sense with a financial instrument.

Please do look over my implementation of the maths. I'm not really a statistician (more in the realm of physics) and so I defer to those who know better than I.

Thanks for the interesting idea,
mashume

 

[ysanchezmd]

Hi mashume,

Thank you for your help, I had changed the code a little, if you can check and give me your opinion I will appreciate, I am not a coder neither a mathematician, just a physician trying to learn trading.

Have great day!

declare lower;

input length = 20;

input nFE = 8;#hint nFE: length for Fractal Energy calculation.
input AlertOn = no;
input Glength = 13;
input betaDev = 8;
input data = close;
def w = (2 * Double.Pi / Glength);
def beta = (1 - Cos(w)) / (Power(1.414, 2.0 / betaDev) - 1 );
def alpha = (-beta + Sqrt(beta * beta + 2 * beta));
def Go = Power(alpha, 4) * open +
4 * (1 – alpha) * Go[1] – 6 * Power( 1 - alpha, 2 ) * Go[2] +
4 * Power( 1 - alpha, 3 ) * Go[3] - Power( 1 - alpha, 4 ) * Go[4];
def Gh = Power(alpha, 4) * high +
4 * (1 – alpha) * Gh[1] – 6 * Power( 1 - alpha, 2 ) * Gh[2] +
4 * Power( 1 - alpha, 3 ) * Gh[3] - Power( 1 - alpha, 4 ) * Gh[4];
def Gl = Power(alpha, 4) * low +
4 * (1 – alpha) * Gl[1] – 6 * Power( 1 - alpha, 2 ) * Gl[2] +
4 * Power( 1 - alpha, 3 ) * Gl[3] - Power( 1 - alpha, 4 ) * Gl[4];
def Gc = Power(alpha, 4) * data +
4 * (1 – alpha) * Gc[1] – 6 * Power( 1 - alpha, 2 ) * Gc[2] +
4 * Power( 1 - alpha, 3 ) * Gc[3] - Power( 1 - alpha, 4 ) * Gc[4];
# Variables:
def o;
def h;
def l;
def c;


# Calculations
o = (Go + Gc[1]) / 2;
h = Max(Gh, Gc[1]);
l = Min(Gl, Gc[1]);
c = (o + h + l + Gc) / 4;


def V = sum(VOLUME, length);
def VWAP = VWap();

def CM3 = fold i = 0 to length with s do s + ((getValue(VOLUME, i) / V) * ((GetValue(C, i) - VWAP)/ 3));
#def CM3 = fold i = 0 to length with s do s + ((getValue(VOLUME, i) / V) * Power((GetValue(CLOSE, i) - GetValue(VWAP, i)), 3));

#def SD = StandardDeviation(length = length);
def sd = StDev(vwap, length);

plot skew = (- CM3 / (SD/ 3));
Skew.AssignValueColor(if skew > 0 then Color.CYAN else if skew < 0 then Color.MAGENTA else Color.YELLOW);

plot zero = 0;

 

[codydog]

I like the ideas - and have started using it yesterday, fwiw. Thanks for the work, cleverness and the community spirit in sharing.

 

[Musk335im3]

This is a very interesting take on the VWAP.

I just went through all my personal trades for this year (winners and losers), and it seems this can help tightening up my entries and exit.

Quick question/ request: is it possible to add an arrow (up/down) whenever the line changes from one color to another (and vice versa). And also maybe a down whenever it crosses the 0 value.

Thanks in advance.

Additional indicator used in this image: MA Crosseover