|
PureBytes Links
Trading Reference Links
|
"the "simple" head and shoulders pattern,"
You know TJ, for someone who is a self-professed ezl beginner, you sure do
have some nads to be saying this. Reminds me a certain recent Texas
Govenor. But I'll give you the benefit of the doubt here and shed some
light on each one of your points nonetheless. First let's start with the
assumption that we're just using EL here.
TJ Sayz:
"ah, bullshit brian :))....here's one...the "simple" head and
> shoulders pattern,"i think even the great french fuzzy programmer said
once that this one
> was difficult with ezl ( and i still haven't seen any reliable code for
> it either!!!)
Brian Sayz:
Head and shoulders is not hard to program in EL or C++ with the right tools.
To do this I would need
1. I could probably do it EL but I reserve the right to use C++. I'm,
assuming a clean feed which your BMI is definitely not. We can get it to
that point with a real time tick filter and decent tick scrubbing
algorithim.
2. Then do something like this:
a. look for most recent hi
b. record it
c. is the next bar another hi?
d. if so replace previous and record new
f if no, then record new low. follow price down.
g. is this bar lower than previous bar
h. if so, record new low
i. if not,preserve existing and check for higher hi
j. if so, we're on the way back up. compare to all time hi
k. are we there yet, are we there yet, are we there yet?
l. if not check for IB and OB
m. if either then preserve existing
n. handle retracement accordingly.
this is extremely crude psuedo code and there would be pitfals i'm sure but
they could be overcome. there's probably a more elegant mathematical
solution that i don't know about but there's someone out there that does,
you can bet on that. you could also have separate congestion algorthims
that look for amount of time spent BELOW the hi. The longer price stays
below altime hi, the more likely a reversal. the point here is that it can
be done and i don't think you're in any position to say one way or the
other. you can increase the sophisitication of your analysis by using a
real language like VC++ and/or VB too.
TJ Sayz:
" i think even the great french fuzzy programmer said once that this one
> was difficult with ezl"
Brian Sayz:
birds of a feature flock together. pierre has even confessed to being
programming illiterate in any other lnaguage except ezl. of course that
doesn't keep him from running off with the mouth sometimes either.
TJ Says:
"here's another one, divergences....i think
> dennis h. said
> that were too many time factors involved with the programming to get it
> right. "
Divergences -- no prob. Let's assume we need to look at several time frames
here so we'll set our code up to scan 5, 10 and 20 bar divergence of an RSI
with price. A lot of the analysis would use the same logic as above where
we follow the indicator point by point and record pivot points. Then we
scan back through that array looking for his and and lows in between the
passing pivot points and corresponging pivot points in price. The user
would define all the time periods they wanted to scan -- 1 to say 100 if
they wanted or just 5, 10, and 20. The software would be configurable that
way. Could you do this with EL -- No but you could with VB or VC and that's
what I said with the right TOOLS and DATA you can do anything.
TJ Says:
so it seems that eyesight viewing charts and human brain interaction are
still the best "systems" for complex patterns.
Brian Sayz:
ah bullshit TJ. if you call those "complex" then you really are a
minimalist trader. trust me there are a helluva lot more complex patterns
out there than H+S and Divergence and they can only be discerned with a PC .
TJ:
"> more complex patterns than those i mentioned above. believe me, i tried
> and spent 10's of thousand of dollars to some of the best programmers i
> could find (incidently, my best programmer was a red chinese
> (now us citizen)
> and he could code rings around the best american coders, but he said that
> new programming language would need to be invented to deal with
> the patterns
> problem)"
B:
When was this? 10 years ago when all we had was C++ and a crude version of
Tradestation that barely would chart and VB was still a glimmer in Gate's
eye? Now I remember Quickbasic was the standard and Applesoft was king or
was it the TRS-80? No wonder your coder came to that conclusion. Now
technology is much improved. You know what, with C++ you CAN invent your
own language to handle any situation!! You can do fu-ing you want IF YOU
KNOW HOW!!! That's the prob, most people don't and most people don't care
to try.
you know I'm tired of cutting and pasting as i read on...rest of my comments
are below.
enjoy your trip to school.
> -----Original Message-----
> From: tradejacker@xxxxxxxxxxxx [mailto:tradejacker@xxxxxxxxxxxx]
> Sent: Wednesday, March 01, 2000 10:01 PM
> To: omega-list@xxxxxxxxxx
> Subject: RE: SURVEY RESULT
>
>
> ah, bullshit brian :))....here's one...the "simple" head and
> shoulders pattern,
> i think even the great french fuzzy programmer said once that this one
> was difficult with ezl ( and i still haven't seen any reliable code for
> it either!!!)....here's another one, divergences....i think
> dennis h. said
> that were too many time factors involved with the programming to get it
> right.
>
> so it seems that eyesight viewing charts and human brain interaction are
> still the best "systems" for complex patterns. you might argue that the
> patterns are visually insignificant and "mind" artifacts, but i disagree.
> the level of programming tools and skills are archaic to
> "systemize" even
> more complex patterns than those i mentioned above. believe me, i tried
> and spent 10's of thousand of dollars to some of the best programmers i
> could find (incidently, my best programmer was a red chinese
> (now us citizen)
> and he could code rings around the best american coders, but he said that
> new programming language would need to be invented to deal with
> the patterns
> problem)
>
> here's another one...try to program whether any single price is a 1 lot
> or a 1000 lot!!!!
In stocks it's called volume and with the right feed (sorry BMI sucks dude!)
this is as simple as retrieving the volume data field on a tick chart in EL
or C++. In futures, you can't do it not because a computer can't but
because THE EXCHANGE DOESN'T GIVE YOU THAT INFO! You can too program it if
the exchange would give you that info. I've never traded on the floow but I
would imgine a floor traders would recognize a 1 from 1000 lot by either
looking at the hand signal and/or volume in the pit.
With regards to your squawk box there are two routes you can go. The ideal
way would be for the guy running the squawk box to input the data into a
feed that was trnasmiited out to your PC. Your software catches that volume
data and presto, your computer can tell the difference. I don't need to
tell you that a PC can't assimilate data if it has no data to assimilate.
The language of the computer is numbers and all input needs to be convereted
into numbers. If you can't do that then we can't begin to talk about this
because like humans, they can't interpret a lnaguage they don't understand.
You're going out on a limb here. But even still here's another solution:
If you don't have then you can take a cruder approach. It involves
translating the squawk box voice data into numbers. Hook up a voice
recognizition software to you damn trading software and have a function that
does nothing but interpret the incoming voice. Use sound filters to reduce
background (white noise) so that only the most prominent (the speaker) voice
is heard then analyze the pattern for numbers. Whenever your function
returns 1 LOT or 1000 LOT then capture that data and write it out to a
database. Record time (down to seconds) and volume. Correlate this with
price and time when you retrieve it. Time is your key field into the
database so that when volume data is coming in and price isn't you can still
match the price (tick) with the volume stamp. Walla, you've got volume
info.
big difference. ya can't program it, but it
> usually leads
> to trends reversals or accelerations and big big profits. a
> mechanical system
> doesn't know the difference while a systematic discretionary trader with
> a squawk would. big big difference and totally UNPROGRAMMABLE!!!!
Nope. Not true. See above. You need to take some progamming classes.
now i'm
> talking about pit trading, which still represents over 92% of the average
> spoo volume vs 8% average e-mini spoo. another one....knowing the buyers
> and sellers, etc......and on and on....a mechanical system doesn't know
> this shit, while i, a systematic discretionary trader, exercises judgment
> and kicks ass with this very important info.
>
> my point is, is there's a whole lot of important crap that can't
> be programmed
> and it's the same ole system trader bigoted arrogance to presume that it
> can be. that is, al least now now...........
Not true my man. Not true. The only assumption I make is that your info
can be assimilated into numbers so the computer can understand it. Besides
who wants to stay glued to your monitor all day when I can have my PC trade
for me. I might be 10% less profit than you on average or I might be 10%
more because my apporach is consistent. Either way, I can go sip pina
colladas on the french riviera and trade many systems at once and you can't.
>
> TJ
>
> and intuition, well, i'll leave that to another time.... :))
Intuition is bullshit too. Intution is stupid because what you're doing is
keying in on something and not even recognizing what that something is. Not
trying to identify exactly what it is you're keying in on and just going by
your gut feeling is, putting it gently, rediculous.
The challenge still stands!
>
> At Wed, 1 Mar 2000 21:47:32 -0800, "Brian Massey" <bnm03@xxxxxxx> wrote:
>
> >In fact, I challenge anybody to name 1 thing that can't be programmed.
> Maybe I'll learn something.
>
>
|