|
PureBytes Links
Trading Reference Links
|
I guess ATR, using three values where STDEV uses one per day, is a lot
quicker to adjust.
Here is a suboptimal GARCH model. The parameters (a0, a1 etc) should
be optimised per market, but you'll get the idea and can play around
with them if you like. It uses the first 100 values to get going.
This is actually an asymmetric GARCH model called a GJR model, and we
like it a lot. But, as it goes with fashion, there are so many
varieties on the theme and lots of zealots for each variety.
{Sample GARCH model}
{MG Ferreira}
{http://www.ferra4models.com}
{For personal use only}
DRet:=C/Ref(C,-1)-1;
SqrRet:=DRet*DRet;
a0:=Stdev(SqrRet,100);
a1:=0.8;
a2:=0.1;
a3:=0.1;
ht:=a0+a1*PREV+a2*Ref(SqrRet,-1)+a3*(Ref(DRet,-1)>0)*Ref(SqrRet,-1);
Sqrt(ht*250);
Regards
MG Ferreira
TsaTsa EOD Programmer and trading model builder
http://www.ferra4models.com
http://fun.ferra4models.com
--- In equismetastock@xxxxxxxxxxxxxxx, "Jose Silva" <josesilva22@xxxx>
wrote:
>
> MG, both of your StdDev solutions still suffer from the same delayed/
> increasing volatility plot problem, which becomes evident on large
> volatility (price gaps).
>
> jose '-)
> http://www.metastocktools.com
>
>
>
> --- In equismetastock@xxxxxxxxxxxxxxx, mgf_za_1999 <no_reply@xxxx>
> wrote:
> > OK, I haven't done this test, not participated in this debate, but
> > to get a *real* feel for the statistical standard deviation, you
> > should calculate it in the percentage change in the close, something
> > like
> >
> > Stdev(C/Ref(C,-1)-1,30);
> >
> > You probably need to also annualise this, something like
> >
> > Stdev(C/Ref(C,-1)-1,30) * Sqrt(250);
> >
> > using a 250 day year. Note that I use 30 days. To get the standard
> > deviation to have any validity (or most other indicators for that
> > matter) you should use 30 or more days. There is a (deep)
> > theoretical reason for this - I think you can maybe take this number
> > down to 20 but do so at your own risk.
> >
> > I also *think* that to do any sensible comparison, you have to use
> > the same number of periods for both. Now, the ATR uses H, L and C,
> > so *maybe* you can use 10 for the ATR when you use 30 for STDEV, but
> > maybe not.
> >
> > Anyhow, looking forward to comments on how this one compares to the
> > ATR.
> >
> > Regards
> > MG Ferreira
> > TsaTsa EOD Programmer and trading model builder
> > http://www.ferra4models.com
> > http://fun.ferra4models.com
> >
> >
> >
> > --- In equismetastock@xxxxxxxxxxxxxxx, "Jose Silva" <josesilva22@xx
> ..> wrote:
> > Manuel, Andrew, staying away from mathematical jargon if possible,
> > let's concentrate on what seems to work best on the markets.
> >
> > Plot and compare these two indicators below any volatile chart:
> >
> > ATR(1);
> >
> > Stdev(C,2);
> >
> >
> > It may be a subtle difference, but I know which one I'd prefer.
> >
> > And introduce Standard deviation to a large price gap over say, 21
> > periods [Stdev(C,21)], and the *increasing volatility* shown by Std
> > Dev *after* the event, is simply wrong.
> > Compare to Mov(ATR(1),21,E).
>
> > Again, from *my own chart observations*, my view is that the ATR is
> > probably the more natural measure of price volatility.
> >
> > My observations and thoughts may not be mathematically correct, but
> > that is the way I view volatility in charts - not as a bunch of
> > abstract numbers to be manipulated mathematically, but rather, data
> > points representing mass psychology at work.
> >
> >
> > jose '-)
> > http://www.metastocktools.com
> >
> >
> >
> > --- In equismetastock@xxxxxxxxxxxxxxx, "Manuel Cabedo"
> > <manelcabedo@xxxx> wrote:
> >
> >> From my own chart observations, I think that the ATR is probably
> >> the best measure of volatility.
> >
> > I don't think so, Jose. Volatility is a kind of dispersion, and the
> > best measure of dispersion is the standard deviation. It is a simple
> > question of statistics. With standard deviation you can do
> > quantitative assertions about the probability of breaking a channel,
> > for instance, or being exited of an operation by a stop.
> >
> > Speaking of securities, I particularly like the standard deviation
> > of daily returns. The distribution of this quantity is not normal,
> > of course, but you can study it on a heuristics base.
> >
> > The work of Bollinger is interesting (I am the translator of his
> > book in Spain) because he always justifies (or tries to.) his
> > methods from a statistical point of view. If someone likes his
> > bands, then reading his book is a must.
> >
> > Once more, thank you, Jose. Your contributions to this forum are
> > always highly valuable (including the one about ATR...).
> >
> >
> > Kind regards.
> >
> > Manuel
------------------------ Yahoo! Groups Sponsor --------------------~-->
In low income neighborhoods, 84% do not own computers.
At Network for Good, help bridge the Digital Divide!
http://us.click.yahoo.com/EpW3eD/3MnJAA/cosFAA/BefplB/TM
--------------------------------------------------------------------~->
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/
|