Keltner Channels

OK, I understand.  I thought there was something more to it and it was providing a trend indicator of some kind.

I'm wondering if there is anyone with the expertise to convert some Tradestation code that I came across to a PCF format that could be represented as a dot indicator in TC2000.

If you have the expertise please help me out with this as I am working on some ideas for MTF style indicators in both TC2000 and TOS. I'm just concerned with the primary formula and not the extraneous parts on how to paint bars, etc.

Here is the Tradestation code:

{ Heikin Ashi PaintBarStudy }
{ for visualization of trend }
inputs: UpColor( RGB(0,255,0)),
DnColor( RGB(255,0,0) ), 
help(”www.TazaTek.com”);
vars: haClose(0),
haOpen(0),
haHigh(0),
haLow(0),
color(0);
if BarNumber = 1 then
begin
haOpen = open;
haClose = (O+H+L+C)/4;
haHigh = MaxList( high, haOpen, haClose);
haLow = MinList( low, haOpen,haClose);
end;
if BarNumber > 1 then
begin
haClose = (O+H+L+C)/4;
haOpen = (haOpen [1] + haClose [1])/2 ;
haHigh = MaxList(High, haOpen, haClose) ;
haLow = MinList(Low, haOpen, haClose) ;
if haClose > haOpen then
color = UpColor
else
color = DnColor;

plotPB(haOpen,haClose,”Ignore-ME-”,color);
SetPlotWidth(1,4);
SetPlotColor(1,color);
end;

diceman or Bruce,   do either of you have the know-how to translate the Tradestation code above into TC2000 PCF? 

I'm revisiting this topic and would really like to understand this original code better.

Thanks in advance.

isitabubble: with the 100 / ABS form you need to create with a customer PCF indicator and NOT a custom % true indicator. That's the only thing I can think of that might be causing you an issue.

IsItPossible:  I'm implementing everything on v16. The formulas you provided simply use longer moving averages but they are not going to get the same result as being able to integrate multiple time frames into a single indicator. And the other solution provided for v7 only  works on the daily time frame anyhow which is not granular enough for what I do.

Therefore, i'm just sticking with having the multiple time frames in watchlist columns. This coupled with the indicator at the bottom of the chart and the ability to toggle different charts using the Function keys gives me everything I need.

I watched my new formula in action with the market today and it is really good. One of the most valuable tools in the 'ol arsenal!

Armed with a better understanding of how Heikin Ahsi closes are constructed I played around with a bunch of different variants tonight and came up with one that I think works quite well.

I'm currently using a 9-period exponential moving average of Heikin Ashi coupled with a 4-period front weighted average of price. I found the that weighted moving average made for more timely signals than calculating the mid point of the range.

Different moving average periods can be substituted with the tradeoff being a slightly earlier entry signal but with more false signals to bump you out of the trend. Longer settings keeps you in the trend longer if you can stomach pullbacks and usually only delays the entry signal by one bar at most.

HA_Trend:  3 WMA & Heikin Ashi  7 EMA

100 / ABS ((3 * C + 2 * C1 + C2) / 6 < (XAVGO7.1 + XAVGH7.1 + XAVGL7.1 + XAVGC7.1) / 4)

HA_Trend:  4 WMA & Heikin Ashi 9 EMA

100 / ABS ((4 * C + 3 * C1 + 2 * C2 + C3) / 10 < (XAVGO9.1 + XAVGH9.1 + XAVGL9.1 + XAVGC9.1) / 4)

HA_Trend:  5 WMA & Heikin Ashi 10 EMA

100 / ABS ((5 * C + 4 * C1 + 3 * C2 + 2 * C3 + C4) / 15 < (XAVGO10.1 + XAVGH10.1 + XAVGL10.1 + XAVGC10.1) / 4)

  HA_Trend:  5 WMA & Heikin Ashi 13 EMA

100 / ABS ((5 * C + 4 * C1 + 3 * C2 + 2 * C3 + C4) / 15 < (XAVGO13.1 + XAVGH13.1 + XAVGL13.1 + XAVGC13) / 4)

I think this finally makes sense to me. 

Which means that my HA_Trend formula (separate thread) is not actually based on a moving average of HA close but instead, an odd invention of my own.

I will have to revisit this and do some more experimenting.

But if HA open is always calculated as (O1 + C1)/2 then it would seem that HA Close would always need to be calculated as (haO + H + L + C) / 4.

Without using haO in the formula you are not carrying forward one of the basic principles that constitute Heikin Ashi.

I&#39;m still not clear on this point as their seems to be a contradiction when it is stated that the HA Open uses yesterday&#39;s open and close, but that data is ignored when calculating the close.

It was my understanding that (O + H + L + C) / 4 is only used to calculate the first bar.

Bruce, i&#39;ve seen it written multiple times now that a Heiken Ashi close is calculated as:

(O + H + L + C) / 4

However, I also understand that the Heiken Ashi open is calculated as:

(O1 + C1) / 2

So wouldn&#39;t that mean the actual Heiken Ashi close should be calculated as:

((O1 + C1) / 2 + H + L + C) / 4    ??

Or as you have stated previously you could substitute out the exponential 3 period moving average of the previous day&#39;s close for the heiken ashi open:

(XAVGC3.1 + H + L + C) / 4

Can you clarify this for me? As I have been using the O1 + C1 / 2 in my Heiken Ashi calculations and would like to understand if that is correct.

Thanks