|
PureBytes Links
Trading Reference Links
|
Alan
I forgot to mention that the weekly formulas (as well as the daily formulas for intraday charts) at
Jose's site all had a bug in the middle 'J' definition. These have now been corrected and for your
information here is the new line.
J:=If(A+LastValue(J)>2 OR B+(Q=1)=2,1,J);
Roy
----- Original Message -----
From: "Roy Larsen" <rlarsen@xxxxxxxxxxxxxx>
To: <equismetastock@xxxxxxxxxxxxxxx>
Sent: Sunday, April 04, 2004 7:45 AM
Subject: Re: [EquisMetaStock Group] Weekly MACD histogram
> Hi Alan
>
> > I have a question on your formula. In some cases, you repeat the
> > definition of a variable. For example:
> > I:=Fml("Calendar Week counter");
> > I:=Abs(I-ValueWhen(2-G,1,I)); M:=G OR I>0;
> >
> > Is this similar to usage in computer programming? For example,
> > x:=x+1. Here, the old value of "x" is being operated on to produce a
> > new value.
>
> Spot on. I find it is often simpler to set up a basic variable and then adjust (or modify) it a
> couple of times than try and include all the logic in only one line. Each succeeding line starts
> with the variable definition already create it and adds another logical component (or not) as
> required.
>
> You can also create an entirely new variable definition using a previously defined variable name.
> The reason for doing this is ususally to get around the 20 var limit, and the new definition would
> be created once the old definition was no longer required for the remaining part of the formula.
>
> > Here are my first attempts to decipher your "Weekly MACD - Close":
> > total confusion and disarray.
> >
> >
> > G:=LastValue(Highest(Sum(DayOfWeek()=F,5))=5);
>
> The purpose of this variable is to identify if the chart is in daily or weekly periodicy. My
> definition of a weekly chart is crude but reasonably effective - if 5 successive bars anywhere in
> the chart are Friday's then consider the chart to be weekly periodicy, otherwise assume daily
> periodicy.
>
> By summing the number of successive Fridays in any 5 bars I know that a result of 5 means that
five
> contiguous bars are Friday. It would be extremely unusual for this to happen on a daily chart and
> almost inevitable on a weekly chart.
>
> Using Highest() means that I only need to see this value once to come to the conclusion that the
> chart is weekly. Using LastValue() makes a logical TRUE (ONE) available for use whereever I want.
> One of the places I use the is in ValueWhen(2-G,J,C) for example. What this says is if J is TRUE
> (greater than or equal ONE) then look back to the previous value of CLOSE, but if G is TRUE then
use
> the current value of CLOSE.
>
> Sorry that my description is hurried but hope this helps.
>
> Roy
>
> > Split into smaller pieces:
> >
> > a:=Sum(DayOfWeek()=F,5);
> > b:=Highest(a);
> > c:=LastValue(b=5)
> >
> > Explanation:
> >
> > a:=Sum(DayOfWeek()=F,5);
> > --------------------------------------
> > The value of F is input by the user. Assume F=5, so no trading on
> > Saturday or Sunday.
> > Assume that today is Monday (=1).
> > Therefore:
> > Today: [DayOfWeek()=5] is False, or 0
> > Previous day, Friday (because F=5): [] is True, or 1
> > Thursday: [] is False, or 0
> > and so on.
> >
> > Therefore,
> > a:= the sum for five periods, (0,1,0,0,0), which is 1
> > It seems to me that the sum will always be "1", so already I must be
> > losing the thread :-)
> > If I plot this bit as an indicator, I get a horizontal line at Y=5
> > going back to 1994. This implies that the daily values are all "1"
> > :-)
> > --------------------------------------
> > b:=Highest(a);
> > --------------------------------------
> > b is the highest ever value for "a", starting from the first day
> > loaded in the chart until today.
> > By my analysis (paralysis?), this can only ever be "1". When plotted
> > in Metastock, it's "5".
> > --------------------------------------
> > c:=LastValue(b=5)
> > --------------------------------------
> > If the last value of the data array "b" is equal to "5", then c:=1
> > (True).
> > Otherwise, c:=0 (False)
> >
> >
> >
> > A few hints would be appreciated.
> >
> > Regards,
> > Alan
> >
> > --- In equismetastock@xxxxxxxxxxxxxxx, "Roy Larsen" <rlarsen@xxxx>
> > wrote:
> > > Hi Alan
> > >
> > > > I made my post at 2233 SHT (Scottish Highlands Time) and your
> > reply
> > > > was at 2350.
> > >
> > > As you can see I don't always reply that quickly, and sometimes I
> > forget to reply at all.
> > >
> > > > At that rate of output (five A4 pages), I estimate that
> > > > your book will take about four days to write. That's a lot less
> > than
> > > > the several weeks it will take me to work through your response
> > point
> > > > by point.
> > >
> > > If only writing a book were that easy :-)
> > >
> > > > I'm amazed at your offer to an obvious know-nothing to "walk me
> > > > through" something that I described as "fiendishly complicated".
> > >
> > > It's not a problem for me.
> > >
> > > > On the subject of probabilities, no statistician would believe
> > that,
> > > > out of the world's total population of 7,000 (according to Fred
> > > > Flintstone -it's probably double that by now), I find myself
> > casually
> > > > boasting to the author about a "fiendishly complicated" formula.
> > If
> > > > the newsgroup moderators had any human feelings at all, they would
> > > > have intercepted your post to prevent the affront to my dignity.
> > >
> > > Your opinion is probably shared by many, but you just happened to
> > be the one that put it in writing.
> > > No offence taken.
> > >
> > > What you need to watch for is when you accidentally post a private
> > reply to a large group. Now that
> > > can be embarrassing.
> > >
> > > Regards
> > >
> > > Roy
> >
> >
> >
> >
> > Yahoo! Groups Links
> >
> >
> >
> >
> >
> >
>
>
>
>
>
>
> Yahoo! Groups Links
>
>
>
>
>
>
Yahoo! Groups Links
<*> To visit your group on the web, go to:
http://groups.yahoo.com/group/equismetastock/
<*> To unsubscribe from this group, send an email to:
equismetastock-unsubscribe@xxxxxxxxxxxxxxx
<*> Your use of Yahoo! Groups is subject to:
http://docs.yahoo.com/info/terms/
|