[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[Metastockusers] Re: How to calculate this ?



PureBytes Links

Trading Reference Links

--- In Metastockusers@xxxxxxxxxxxxxxx, "Karile" <karile@xxxx> wrote:
> 
> Hi Kut2k2,
> 
> Thanks for your reply.
> 
> As you said, I think it is going to be difficult to implement in 
Metastock.
> The formula you gave to me is different.
> 
> The main purpose is to classify (ascending order) a time series (ie 
: 100
> values) and then pick the 50th(x) and 51th(y) values.
> 
> Then, the result (median) would be :(x+y)/2
> 
> The median returns a number which divides in two the sample (values 
above
> it/below it).
> 
> Thanks for your help,
> 
> Regards,
> 
> Karile

Hi again, Karile.

Sometimes I'm too subtle for my own good. What I should have said was 
that a 100-bar median filter would be a lot of work for very little 
gain (if any) over an ordinary 100-bar MA. FWIW I personally would use 
a 3-bar median filter followed by an MA of the length you desire. And 
you might want to ask yourself what's so special about '100'. Are you 
just copying what some trading "guru" recommended without really 
thinking about whether it makes sense for your needs? One final piece 
of trading advice I can offer you: Question everything you read and 
hear.

HTH,
kut2k2

> ----- Original Message -----
> From: "kut2k2" <kut2k2@xxxx>
> To: <Metastockusers@xxxxxxxxxxxxxxx>
> Sent: Tuesday, May 06, 2003 6:44 PM
> Subject: [Metastockusers] Re: How to calculate this ?
> 
> > --- In Metastockusers@xxxxxxxxxxxxxxx, "Karile" <karile@xxxx> 
wrote:
> > >
> > > Hi,
> > >
> > > I would like to know if it is possible with Mestatock to 
calculate a
> > median
> > > (statistically speaking) :
> > >
> > > The median is the middle number (value) in any given set of 
numbers
> > (values)
> > > :
> > >
> > > For example :
> > >
> > > A set of values :
> > >
> > >  20 45 10 3 90 19
> > >
> > > 1) We have to classify (rising order) these values : so
> > >
> > > 3 10 19 20 45 90
> > >
> > > 2) Because we have an even set of numbers ( 6 values) :
> > >
> > > We take the value 19 and the value 20 :
> > >
> > > (19+20)/2 = 19.5 = MEDIAN
> > >
> > > 3) for an odd set of numbers
> > >
> > > Let's take the same set of numbers but with 5 values (odd 
number) :
> > >
> > > 20 45 10 3 90
> > >
> > > We classify :
> > >
> > > 3 10 20 45 90
> > >
> > > MEDIAN = 20 (the number which divides in two the set of numbers)
> > >
> > >
> > > Is it possible to do the same thing thing with Metastock (I 
would
> > like to do
> > > it with 100 values) ?
> > >
> > > Any help appreciated,
> > >
> > > Regards,
> > >
> > > Karile
> >
> > Hi, Karile
> >
> > I don't know what exactly your intention here is of course but if 
your
> > 100 values represent the lookback period of a time series, you're
> > probably defeating your purpose. A 100-bar median filter would 
have
> > approximately the same result as a 100-bar moving average with the
> > same disadvantage: about 50 bars of lag.
> >
> > I use a 3-bar median filter as a "prefilter" or "preprocessor" 
when I
> > make indicators of price velocity in order to "snap off the 
jaggies"
> > because velocity is noisier than raw price. The cost is only one 
bar
> > of lag, but that is more than paid back by being able to use a 
shorter
> > MA on the smoother output of the prefilter.
> >
> > There is a general-purpose median algorithm somewhere on the Net 
but
> > it is pretty complicated and probably involves looping.  For short
> > medians, there are more efficient algorithms.
> >
> > For the 3-bar median filter:
> >
> > Median := Max(Max(Min(P, Ref(P,-1)),Min(P, Ref(P,-2))), 
Min(Ref(P,-1),
> > Ref(P,-2)));
> >
> > HTH,
> > kut2k2



------------------------ Yahoo! Groups Sponsor ---------------------~-->
Rent DVDs from home.
Over 14,500 titles. Free Shipping
& No Late Fees. Try Netflix for FREE!
http://us.click.yahoo.com/BVVfoB/hP.FAA/uetFAA/zMEolB/TM
---------------------------------------------------------------------~->

To unsubscribe from this group, send an email to:
Metastockusers-unsubscribe@xxxxxxxxxxx

 

Your use of Yahoo! Groups is subject to http://docs.yahoo.com/info/terms/