|
PureBytes Links
Trading Reference Links
|
I sincerely appreciate your reply, Brian; however, there is significantly
more to pattern recognition software than static code. The syntax must be
adaptable to pattern distributions which fall outside of 'normative
mathmatical' values.
In your description of a parabola, there are unique and numerus variations
of this structure that mandate certain 'neural' coding innovations, none of
which I have had the pleasure of isolating, at any cost thus far in my
pursuits.
Perhaps you have a mathmatical model which is operative in this regard?
I have isolated ther patterns that are far more effective than the C&H, and
have worked for 5+ years to code them, in any language... No luck!
Rick Martinelli, of Haiku Laboratories, right here in my backyard, has
programmed the cup and handle pattern, as noted in a previous edition of
S&C; however, the conversion into a trading platform always seems to be the
roadblock, ad nauseaum.
Perhaps the most intriguing software is the french fuzzy logic so disparaged
on this list, yet very few intelligible threads have followed up on this. I
think its about time I ventured over that way...
Many Thanx,
Lee
----- Original Message -----
From: Brian Massey <bnm03@xxxxxxx>
To: spi <spi@xxxxxxxx>
Sent: Friday, March 10, 2000 6:15 AM
Subject: RE: STK - chart pattern to die for
> "> how about a bell shaped curve and its derivatives?"
>
> This is one way. And I'm sure there are so-called patterns out there that
> are harder to program than others, but the overwhelming maajority aren't
> IMO. My point to this entire discussion was that anything can be
programmed
> if you can define it. If you can't define it why are you trading it? I
> garantee there are better ways to trade than using some hit or miss chart
> pattern that can't even be quantitied. The problem with visual trading is
> that it's not defineable and this is pricesely why most people continue to
> try to trade visually -- because they can't define, they can't test a
> pattern they think is valid, and of course they're always right.
>
> The thing about the markets is time frame. Right? I mean if you wanted to
> approximate, say using TJ's example, the rounded bottom of a suacer as it
> was setting up, you would have to define the lookback period (where to
> start). Are we looking for setups the encompass 20 bars, 40, bars, or 100
> bars? The calculated values for each of these timeframes will be
different
> and so will the analysis. While the code would be long it could be done.
> And once it was written, it would save manhours. But if you can't define
a
> pattern then how can you backtest it and trade it? If you can't quanitfy
> then I would argue that not only is the pattern suspect in terms of it's
> reliability, but also that there are better, more quantifiable ways to
> trade. By definition, a pattern is somethig that regularaly repeats. If
it
> doesn't regularly repreat then it's not a pattern. I think a lot of
people
> like TJ think they see a pattern when really they don't. And since they
> can't backtest it, all we have to go on that it works is their aggressive
> opinions of wheather they're right or not. They can't prove it.
>
> PS I think they have a cup with handle code on the Omega website, in EL
> noless.
>
> > -----Original Message-----
> > From: spi [mailto:spi@xxxxxxxx]
> > Sent: Thursday, March 09, 2000 9:37 PM
> > To: Brian Massey
> > Subject: Re: STK - chart pattern to die for
> >
> >
> > how about a bell shaped curve and its derivatives? "no problem?" had the
> > 'best' minds in the business... pattern recognition is far from easy...
> > lee
> > ----- Original Message -----
> > From: Brian Massey <bnm03@xxxxxxx>
> > To: List, Omega <omega-list@xxxxxxxxxx>
> > Sent: Tuesday, March 07, 2000 6:19 PM
> > Subject: RE: STK - chart pattern to die for
> >
> >
> > > Come up with a solid definition and no problem. It's just a
> > parabola with
> > a
> > > vertex, x and y axis and it meets resistance on the upswing. The
handle
> > is
> > > a retracement before takeoff. This should be accompanied by
> > larger volume
> > > of course as players come into support price. The distance from the
> > bottom
> > > of the cup to the top should give you an inital profit target. These
are
> > all
> > > definable so I don't see what the big deal is. For all of you
> > feely-touchy
> > > types out there, anything less is boarderline-gambling. You can
> > look at it
> > > from every single timeframe you want. This gets kind of advanced and
> > Tough
> > > S--t 2000 might choke on the load. But with the right tools it can be
> > done.
> > >
> >
>
|