Support / Resistance using VAP

[JeanChristophe]

Hello,
I am exploring the idea of getting support/resistance from volume (not really new one, I do agree).

Inspired from Volume at Price --> Explore Possible? and I ended up with this code which:
compute VAP for the last 63 days,
extract the 4 biggest volume peak
get the associated price
plot the price corresponding to the volume peak for each and every day.

Being new to AFL, I am not sure about the strategy I used. Feedback on improvement would be appreciated.
Thanks in advance.

priceH = IIf( H == L, H+1e-5, H); // to avoid error if H = L , ref https://forum.amibroker.com/t/volume-at-price-explore-possible/8910/2

lookback = 63;



for ( i = lookback; i < BarCount; i++ )
    {
	//Compute VAP
	mx = PriceVolDistribution( priceH, L, V, bins = 100, true, i-lookback, i );
	row_size = 100; 
	
	//Get Volume & Price Distribution
	vol_distrib = MxGetBlock(mx, 0, row_size-1, 1,1,True);
	price_distrib = MxGetBlock(mx, 0, row_size-1, 0,0,True);
	
	//Find Volume peaks using PEAK function
	pk = Peak(vol_distrib, 30, 1);
	pk = IIf(IsNull(pk), 0, pk);
	pos = Ref(pk,-1) != pk;
	
	//Create a matrix with volume 0 if no peak
	mx_peak = Matrix( row_size, 2, 0 );
	mx_peak = MxSetBlock(mx_peak, 0, row_size-1, 0,0,price_distrib);
	volume_peak = vol_distrib * pos;
	mx_peak = MxSetBlock(mx_peak, 0, row_size-1, 1,1,volume_peak);
	
	//Sort Matrix to access corresponding price values
	mx_peak_sorted = MxSortRows(mx_peak, False, 1);
	
	//Extract the first 4 biggest volume
	sl_1[i]=mx_peak_sorted[0][0];
	sl_2[i]=mx_peak_sorted[1][0];;
	sl_3[i]=mx_peak_sorted[2][0];;
	sl_4[i]=mx_peak_sorted[3][0];;
	
    }
    
 Plot(sl_1, "sl_1", colorRed, styleLine | styleThick);
 Plot(sl_2, "sl_1", colorBlue, styleLine | styleThick);
 Plot(sl_3, "sl_1", colorGreen, styleLine | styleThick);
 //Plot(sl_4, "sl_1", colorBlack, styleLine | styleThick);

[sinecospi]

Hello @JeanChristophe

I've created a new AFL on my PC with your code, and it takes ~3sec for the screen to refresh!, and when I look at "Code check & Profile", there are no errors, however, there are ~18,000 executions of Peak() on ~18,000 EoD bars (S&P 500 Index since 1950):

In your post:

@JeanChristophe

compute VAP for the last 63 days,

I suspect that the starting point of the for() loop needs modification:

sl_1	= sl_2	= sl_3	= sl_4	= Null ;

cnsStartBar	= Max(0, BarCount - lookback) ;   // Protection against an array out of bounds error

//for ( i = lookback; i < BarCount; i++ )
for ( i = cnsStartBar; i < BarCount; i++ )

After the mod, it now completes in ~6 millisec:

You might want to limit the calcs to the visible bars.

As to whether it provides an edge, I'm still evaluating.