|
PureBytes Links
Trading Reference Links
|
I was just checking the Wright model (out of curiosity, I'm not too
familiar with macroeconomics :) and noticed the argument in the "F"
functions perhaps may become negative(?). Just in case then you'd
need to adjust the HAS function because as such it is only valid for
x>0.
Also it is a good idea to cap off possible large values to prevent
useless processing time.
The one below should safely handle both pos/neg and large/small nrs.
This probably solves your problem, if not let me know and give me an
input value that causes trouble. Good luck.
function t(x)
{ return 1 / (1+0.2316419*x); }
function HASpre(x)
{ return 1 - (exp(-0.5*x^2)/sqrt(2*3.141592654))*(0.31938153*t(x)-
0.356563782*t(x)^2+1.781477937*t(x)^3-1.821255978*t(x)
^4+1.330274429*t(x)^5); }
function HAS(x)
{ return IIf(x<-6, 0, IIf(x>6, 1, IIf( x>0, HASpre(x), 1 - HASpre(-
x)))) ; }
-treliff
--- In amibroker@xxxxxxxxxxxxxxx, "sebastiandanconia"
<sebastiandanconia@xxx> wrote:
>
>
> Treliff, am I doing something wrong or does HAS have a weakness with
> limits? NORMSDIST() only generates probabilities between 0% and
100%,
> but HAS is generating numbers outside of that range and some of
them are
> clearly wrong. (Mathematically correct, I'm sure, but clearly
wrong in
> terms of what it's supposed to be indicating. )
>
>
>
> Luck,
>
> Sebastian
>
>
>
>
> --- In amibroker@xxxxxxxxxxxxxxx, "treliff" <treliff@> wrote:
> >
> > You're welcome. Rest assured I didn't figure out the numbers
myself :)
> >
> > Essence is that NORMSDIST is an integral that cannot be coded in
AB.
> > HAS is a polynomial that does a very good approximation.
> >
> > A while ago I submitted to TJ suggestion to include NORMSDIST and
> > NORMSINV (its inverse) as native AB functions. These are essential
> > functions for any statistical approach.
> >
> > I don't know how Tomasz would code 'm, but in any case it would
speed
> > up things, I assume. So far no luck.
> >
> > So Sebastian, or anyone using these functions, feel free to
endorse
> > this effort by submitting a similar request at the feedback
center.
> >
> > http://www.amibroker.com/feedback/
> >
> > Thanks :)
> >
> > -treliff
> >
> > --- In amibroker@xxxxxxxxxxxxxxx, "sebastiandanconia"
> > sebastiandanconia@ wrote:
> > >
> > >
> > > A sincere thank-you for the response, I know it's a pretty
arcane
> > > subject. This is the nearly unintelligible math part I was
talking
> > > about when I said in a subsequent post that I didn't realize how
> > much I
> > > was asking for, LOL! I don't understand this enough to apply it
to
> > > what I want to do. I assume it's just a formula where I plug in
the
> > > values, but the "where", "how" and "why" plumb evade me.:) I'll
> > work
> > > on it, though.
> > >
> > >
> > >
> > > Luck
> > >
> > > Sebastian
> > >
> > >
> > >
> > > --- In amibroker@xxxxxxxxxxxxxxx, "treliff" <treliff@> wrote:
> > > >
> > > > Sebastian,
> > > >
> > > > The HAS function below (based on Hasting's polynomial
> > approximation)
> > > > gives 7-decimal accuracy of NORMSDIST.
> > > >
> > > > function t(x)
> > > > { return 1 / (1+0.2316419*x); }
> > > >
> > > > function HAS(x)
> > > > { return 1 - (exp(-0.5*x^2)/sqrt(2*3.141592654))*(0.31938153*t
(x)-
> > > > 0.356563782*t(x)^2+1.781477937*t(x)^3-1.821255978*t(x)
> > > > ^4+1.330274429*t(x)^5); }
> > > >
> > > > -treliff
> > > >
> > > >
> > > > --- In amibroker@xxxxxxxxxxxxxxx, "sebastiandanconia"
> > > > sebastiandanconia@ wrote:
> > > > >
> > > > >
> > > > > I came across what I consider to be a valuable stock
> > market/economic
> > > > > indicator, the Wright Model "B" yield-curve indicator. Using
> > this
> > > > > formula in Excel:
> > > > >
> > > > >
> > > > > Probability = NORMSDIST(-2.17 - 0.76 x S + 0.35 x R)
> > > > >
> > > > > where "S" is the spread (10-Year Treasury yield minus 3-
month T-
> > Bill
> > > > > yield) and "R" is the Fed Funds rate, it gives the
probability
> > of
> > > > > economic recession within the next 4 quarters. (Only about
44%
> > > > right
> > > > > now, so there's some good news. I envision using this as a
> > market-
> > > > exit
> > > > > indicator, warning when conditions are about to turn really
> > ugly for
> > > > > both the stock market and the economy. )
> > > > >
> > > > > This formula:
> > > > >
> > > > >
> > > > > Z(x) = (1/(sqrt(2*pi()))*exp(-x^2/2))
> > > > >
> > > > > appears to be the actual math represented by the NORMSDIST
> > > > function. I
> > > > > believe AB supports all the operations in this formula.
> > > > >
> > > > > My problem is that I'm not math-savvy enough to make the
leap
> > from
> > > > here
> > > > > to turn this into a complete AB formula. I don't know what
> > > > operation
> > > > > the NORMSDIST formula performs on the Wright Model part, I
don't
> > > > know
> > > > > what the "x" variable is supposed to be...there's no end to
> > what I
> > > > don't
> > > > > know.:)
> > > > >
> > > > > Any help from my superiors in the math field (undoubtedly a
VERY
> > > > large
> > > > > club) would be greatly appreciated.
> > > > >
> > > > >
> > > > >
> > > > > Luck to all,
> > > > >
> > > > > Sebastian
> > > > >
> > > >
> > >
> >
>
Content-Description: "AVG certification"
No virus found in this incoming message.
Checked by AVG Free Edition.
Version: 7.5.441 / Virus Database: 268.17.34/679 - Release Date: 2/10/2007 4:08 PM
|