Re: LinRegSlope

[tradeshark@xxxx]

Hi David,

here are 2 MetaStock formula for LinRegslope and some additional info.

Info has been taken from :

http://www.guppytraders.com/gup98.htm

best source for MS formula on the web - as far as I know.

hope it helps ;-) - regards
Tradeshark

1. Lb:=Input("Look-Back Periods?",3,100,13);
ROC( LinRegSlope(C,Lb),Lb,$)


2nd. :

{The basic "programming" in MetaStock of the LRS-ROC indicator is 
like this:
NIO is the _basic_ number of points taken to calculate the ROCs. The 
four 
ROCs _per day_ used for interpolation are:}

rll:=ROC(O,nio-1,%)/(nio-1);
rl:=ROC(O,nio,%)/nio;
rh:=ROC(O,nio+1,%)/(nio+1);
rhh:=ROC(O,nio+2,%)/(nio+2);

{Now, interpolation is done at a point XIO (0<=xio<=1), i.e., between 
the
points NIO and NIO+1, using a simple interpolation formula weighting 
with
the _distances_ of the four ROCs from the interpolation point XIO:}

rit:=(rll/(1+xio)+rl/(xio+.0001)+rh/(1-xio)+rhh/(2-xio))/(1/(1+xio)+1/
(xio+.
0001)+1/(1-xio)+1/(2-xio));

{Then the LinearRegressionSlope is taken _twice_ (for numerical 
reasons)
from the interpolated ROC, using the basic number of points; i.e., 
NIO. That
is where we get our LRS-ROC indicator, named RO here:}

ro:=LinRegSlope(LinRegSlope(rit,nio),nio);

{NIO+XIO together represent the _real_ (not integer!) number of 
points or
days, the LRS-ROC indicator is calculated for, e.g. 4.57 days.
Unfortunately, MetaStock is not able to use _one_ real number (e.g., 
4.57)
as input to be splitted into NIO and XIO and to be used in the 
formulas
above. So NIO and XIO have to be provided separately, because of the
restrictions in MetaStock, and both are subjected to 
some "optimization" in
accordance to the historical prices.

To be clear: This LRS-ROC indicator triggers _only the timing_ for
entering/closing a position, using an appropriate criterion. 
Additional
(also ROC-based) criterions are used to stay out/in during extreme
bearish/bullish situations.

Additionally: This "TA" is only one basic element of my option trading
system, primarily to catch some "special reality effects" that cannot 
be
modeled by example-based know-how recycling from historical data. But
probably this TA system may also be used as a "stand-alone" system.

rudolf stricker





--- In amibroker@xxxx, "David Holzgrefe" <dtholz@xxxx> wrote:
> Hi All I was trying to find an explanation of the Linregslope how 
its calculated and the best methods to use it 
> 
> maybe someone could point me to a URL that would explain it in more 
depth
> 
> Thanks David
> LinRegSlope