Re: Reverse-engineering Time Segmented Volume

[countachl <countachl@xxxxxxxxxxx>]

> Subject: Reverse-engineering Time Segmented Volume Indicator
> Date: Wed, 11 Feb 2004 20:03:18 -0500
> From: "CRE -- Earthlink" ?rengholm@xxxxxxxxxxxxx?
> To: ?omega-list@xxxxxxxxxx?
>
> I use both Tradestation and Worden's TC2000. The proprietary Time Segmented
> Volume oscillator in TC2000 seems to work quite similarly to On Balance
> Volume; however, there are a number of differences. I'm trying to
> reverse-enginner the TSV indicator so I can incorporate it into a system
> that can be backtested in Tradestation. Has anyone ever done this, and do
> you have either the TS code or the algorithm that you might be willing to
> share?
>
> With OBV, the entire day's volume is either added to, or subtracted from a
> running total depending upon whether the price was up or down -- regardless
> of how much the price rose or fell. TSV on the other hand takes both price
> and volume amounts into account. For example, in TC2000, if you set TSV to a
> period of 1, it looks like the TSV can be approximated as:  volume *
> (percentage price change since last bar) * (a scaling factor)
>
> However, it is not clear to me how the running accumulation takes place when
> TSV is set to a longer length. The daily increment does not appear to be
> just added to or subtracted from a running total, nor is a moving average of
> the daily increment.
>
> Any ideas?  Thanks.
>
> Rudy
> rengholm@xxxxxxxxxxxxx
>

TSV  basically seems to be OBV with a length parameter.
I coded it that way several years ago.
It has very high correlation with the Worden TSV.

Here is a discussion from the TradestationWorld site:

https://www.tradestationworld.com/discussions/Topic_Archive.aspx?Topic_ID=2397

"The OBV% has the same general shape of the TSV, but is much noisier,
and has more directional variations than the OBV_L code. It does use a length
parameter, and is referenced to 0.

The OBV Multivote code produces a line almost identical to the OBV.
There are some numerical differences, but always the same shape and direction.
Like two moving averages that are very close in length.

My OBV_L indicator has almost the same shape, but some
0-crossings are displaced a few bars, and the amplitude may be different.
It is very close, though. Much closer than OBV%.

Here is the OBV_L code:

{Indicator - OBV_L (OBV with Length ) }

input: Len(20);
var: TodayV(0), OBV_L(0) ;

IF C >= C[1] Then TodayV = Volume ;
IF C < C[1] Then TodayV = Volume * -1 ;

OBV_L = Summation(TodayV,Len)/1000 ;

Plot1 (OBV_L, "OBVL") ;
Plot2 (0,"Obv0") ;
==================================================