|
PureBytes Links
Trading Reference Links
|
Actually you made a good point. Eckhardt did write an article warning
of problems with indicators that fail the "c-test." Which basically
means that if a change in the unit of measure changes an angle or
indicator value, that approach will give inconsistent results like
when a stock splits. To do the c-test in AB "split" the data. If the
indicator or angle is the same both pre and post split, it passes the
c-test. If not it fails. If anyone wants a pdf of the article, email
me privately. P.S. Linregslope applied directly to a price array fails
the c-test. Which doesn't mean that linregslope is worthless BTW.
--- In amibroker@xxxxxxxxxxxxxxx, "sebastiandanconia"
<sebastiandanconia@xxx> wrote:
>
> In one respect, I shot my mouth off before fully understanding that
> drawing lines using linregslope isn't the same as simply drawing lines
> "freehand," whether it's with pencil and paper or with computer
> graphics. Sincere apologies.:)
>
> There's still the issue of how important the angle really is. Even if
> using linregslope yields angles that are consistent without regard to
> scaling, that's interesting from a trig standpoint but from a trading
> standpoint, so what?
>
>
> Luck,
>
> Sebastian
>
>
> ---- In amibroker@xxxxxxxxxxxxxxx, "Fred" <ftonetti@> wrote:
> >
> > Yes ...
> >
> > The angle is a relation of rise ( Price ) over run ( Time )
> >
> > When rise is measured in terms of percent or log10 of price there
> > will be no change in the angle after a split ... Percentage or log10
> > based moves in price are really the only reasonable way to measure
> > them any way ... or chart them for that matter ... this is for the
> > same reason that CAR is calculated the way it is i.e Total Gain ^ (
> > 1 / Number of Years ) as opposed to something that typically gets
> > referred to as ANN and has a formula like Total Gain / Number of
> > Years ... The latter is at least imho a semi meaningless statistic.
> >
> > --- In amibroker@xxxxxxxxxxxxxxx, "quanttrader714"
> > <quanttrader714@> wrote:
> > >
> > > Would you agree there's a problem if the slope changes when the
> > stock
> > > splits?
> > >
> > > --- In amibroker@xxxxxxxxxxxxxxx, "Joe Landry" <jelandry@> wrote:
> > > >
> > > > Don't apologize but you're not bursting my bubble, yet! ( but I'm
> > > open to
> > > > having it deflated).
> > > > I've heard this knock off before and let's put it to bed if we
> > can.
> > > >
> > > > Did you try zooming the chart? Both x and y? Surely the slope
> > of the
> > > > plotted line
> > > > changes as you move both the x, abscissa and the ordinate, price,
> > > but the
> > > > calculated
> > > > values of the slope from using linregslope does not. I think of
> > the
> > > slope
> > > > as representative of the rate of change of
> > > > that price (or other) array. I don't know what Eckhardt said or
> > in
> > > what
> > > > context he said it. Maybe
> > > > he was talking about using pencil and paper. There the scaling on
> > > the chart
> > > > would make a difference.
> > > >
> > > > Also, you have to ask yourself, why would Tomasz have coded a
> > > linregslope
> > > > function, or Dimitris Tsokakis and others
> > > > used it so frequently in their work on the forum? In my
> > collection
> > > alone of
> > > > valued AFL clips I get 200 hits by many different
> > > > users of this board.
> > > >
> > > > Best regards
> > > > JOE
> > > >
> > > >
> > > > ----- Original Message -----
> > > > From: "sebastiandanconia" <sebastiandanconia@>
> > > > To: <amibroker@xxxxxxxxxxxxxxx>
> > > > Sent: Thursday, August 24, 2006 9:31 AM
> > > > Subject: [amibroker] Re: Never Took Trigonometry
> > > >
> > > >
> > > > >I apologize for bursting your bubble, but angles are not going
> > to be a
> > > > > consistent measure. If the scale of the chart changes so does
> > the
> > > > > angle, even if the price data and timeframe are precisely the
> > same.
> > > > >
> > > > > Trendlines that connect highs/lows, however, are consistent
> > regardless
> > > > > of scale. In "The New Market Wizards" mathmatician William
> > Eckhardt
> > > > > explains why methods based on angles are fallacious, but
> > methods based
> > > > > on trendlines are more legitimate.
> > > > >
> > > > >
> > > > > Luck,
> > > > >
> > > > > Sebastian
> > > > >
> > > > > --- In amibroker@xxxxxxxxxxxxxxx, "Charles J. Dudek" <trader@>
> > wrote:
> > > > >>
> > > > >> I don't know how to convert a slope value (LinRegSlope) to an
> > angle.
> > > > >> I took a line from the Woodie's CCI script and converted it,
> > but I
> > > > >> don't think it's right.
> > > > >>
> > > > >> PI = atan(1.00) * 4;
> > > > >> angle = round(180 * acos(1/LinRegSlope(C,sp)) / PI);
> > > > >>
> > > > >> Chuck Dudek
> > > > >>
> > > > >
> > > > >
> > > > >
> > > > >
> > > > >
> > > > >
> > > > >
> > > > > Please note that this group is for discussion between users
> > only.
> > > > >
> > > > > To get support from AmiBroker please send an e-mail directly to
> > > > > SUPPORT {at} amibroker.com
> > > > >
> > > > > For other support material please check also:
> > > > > http://www.amibroker.com/support.html
> > > > >
> > > > >
> > > > > Yahoo! Groups Links
> > > > >
> > > > >
> > > > >
> > > > >
> > > > >
> > > > >
> > > > >
> > > >
> > >
> >
>
|