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

Herman: Re: Putting the cart before the horse?



PureBytes Links

Trading Reference Links


Hi Dave,I finally got back to your email ... hope you are still 
around :-) I find your post very interesting and have what are, undoubtedly, too 
many questions. Sorry that this appears a one way street: there isn't much I can 
contribute to this topic.
Herman,<?xml:namespace prefix = 
o ns = "urn:schemas-microsoft-com:office:office" />
My responses to your questions 
are below.  Please don&#8217;t think of me 
as anyone who necessarily knows what he&#8217;s talking about&#8212;I&#8217;m not convinced that I 
do.  Just consider these as inputs, 
another point of view, one person&#8217;s way of looking at things from a possible 
choice of many.At 10:04 AM 7/3/01 -0700, you wrote:>...I 
start by looking at a certain market, and trying to break that up into pieces 
with different>character, e.g. trending, trading range, reversing, etc. 
So, if I'm trying to build a reversing>system using an oscillator(s), 
I'll want to select those segments of the market that are reversing>or 
trading in a range. Assuming I have enough bars, I can break those into segments 
for in and>out of sample testing...Do I understand correctly 
that, given a long trend, you'll test the same trending period in segments? 
Do you use overlapping sample periods?
I&#8217;d rather use successive, long trending periods assuming I 
have enough data and sufficient trades are generated.<SPAN 
style="mso-spacerun: yes">  I&#8217;d rather not use overlapping 
periods.  Again, I&#8217;ll start out with 
some sort of an idea for an entry like, &#8220;I want to buy pullbacks in a 
trend&#8230;&#8221;  This was a popular method 
during the recent bull run.  Does it 
still work?  Perhaps, depending on 
the timeframe you are using.
So, I&#8217;d want to start developing 
a system on some markets that are very trendy.<SPAN 
style="mso-spacerun: yes">  I&#8217;d find those in a scan of my database, 
using some criteria that I would establish to measure trendiness.<SPAN 
style="mso-spacerun: yes">  This should give me some markets 
(securities) which tend to be more trendy than others.<SPAN 
style="mso-spacerun: yes">  Hopefully, it will give me more trends 
& bars to work with.>Once I've found a successful approach 
to the specific character of that market, I can expand my>testing to a 
larger portion of the historical data, but I'll need to find a way to turn the 
system>on and off depending on what type of market behaviour is being 
exhibited. Then I'll run this>overall system against the larger data set, 
in much the same manner as you describe below.How many types of market 
behavior do you recognize, do you consider patterns (i.e. head & shoulders) 
a market behavior? You mentioned Trend, trading range, reversal point, ... any 
others?
For me, I&#8217;m sticking with some 
very straighforward approaches like trend following and volatility breakout, 
because they are fundamentally appealing to me&#8212;I can rationalize why this type 
of a trade will continue or fail.  
H&S patterns or any other type of behaviour that makes sense to you 
is what really matters.  I would 
recommend a system developer focus on those patterns or market actions that are 
the most intuitively appealing.Do you use exporations or other 
methods to classify the market?
I use Explorations to find those 
markets that I&#8217;ll use in building and refining my system.<SPAN 
style="mso-spacerun: yes">  I suppose this is a way of classifying 
markets and selecting those which match.>...A personal belief 
of mine is that market behavoiurs are persistent. The S&P is much more 
likely>to continue it's reversing behaviour in the future. I'm fairly 
confident that next month, it>won't become a trendy market.I am 
intrigued by the idea of classifying stocks by their behavior. I tend to assign 
as much value to a system that works as to a system that doesn't work, they both 
carry an important message for us to decode. I 
agree completely.  They both offer 
lots of information.  The classic 
example is the system that loses 90% of the time, and loses by a lot.<SPAN 
style="mso-spacerun: yes">  I wonder if I just reversed my 
entries&#8230;.Did you ever make an attempt to develop a classification 
system for stocks? I don't mean classifications by 'temporary' characteristics 
like trending, I mean more permanent, what appear like price-independent 
characteristics. An example would be a security's sensitivity to certain 
oscillators.I haven&#8217;t.<SPAN 
style="mso-spacerun: yes">  I&#8217;m using these periods of trending and 
flat to build my systems.  In the 
end when running across a portfolio, I want the system to make that 
determination.
Since Price and Volume are the 
only two pieces of data we have, all indicators are derived from one, the other, 
or a combination. My approach is that the indicator is just an aid to seeing the 
price action more easily.  

Most indicators are correlated 
because they are all based on price in one way or another.<SPAN 
style="mso-spacerun: yes">  Adding too many to a system just 
engenders curve fitting and unrealistically inflates results.
One technique that can be very 
beneficial is to focus on making your portfolio or testing folder less 
correlated.  If you had all bank 
stocks, plus some brokerages, you are probably not getting a lot more 
information from your system tests than just using the Banking or Brokerage 
indicies.>A sidenote here is that I'm making certain assumptions 
about basic system design. I subscribe>heavily to the signal / trigger 
approach as well as Chuck LeBeau's methods for developing robust>systems. 
(www.traderclub.com) I really like adaptive rather than fixed approaches. Also, 
you'll>notice that markets look very different depending on the 
timeframe: comparing a 3 minute chart of>the S&P with a daily chart 
is a good example.Chuck's bulletins are many ... I don't recall him 
referring specifically to robustness. Are you referring to his overall approach 
or to specific measures designed to measure and improve robustness?<SPAN 
style="COLOR: #333399">I&#8217;m referring to his overall approach.<SPAN 
style="mso-spacerun: yes">  I&#8217;ve read and reread all of his 
bulletins numerous times, as well as his book.<SPAN 
style="mso-spacerun: yes">   All are excellent.<SPAN 
style="mso-spacerun: yes">  The chandelier exit using a multiple of 
ATR() is one example of a robust, widely applicable exit technique.<SPAN 
style="mso-spacerun: yes">  I also like his method of taking profits 
(or partial profits) at an extreme ATR() move in a positive 
direction.When you say "adaptive" do you refer to a human or a 
formulae quality?<SPAN 
style="COLOR: #333399; FONT-FAMILY: 'Times New Roman'; FONT-SIZE: 12pt; mso-fareast-font-family: 'Times New Roman'; mso-ansi-language: EN-US; mso-fareast-language: EN-US; mso-bidi-language: AR-SA">Formulae.<SPAN 
style="mso-spacerun: yes">  In my systems, I&#8217;m counting on the human 
behaviour to remain consistent and predictable (i.e. when the S&P breaks a 
pivot to the upside, traders sell the rally, and the market pulls back before it 
continues upward.)<SPAN 
style="FONT-FAMILY: 'Times New Roman'; FONT-SIZE: 12pt; mso-fareast-font-family: 'Times New Roman'; mso-ansi-language: EN-US; mso-fareast-language: EN-US; mso-bidi-language: AR-SA">>Just 
some somewhat random ideas to go with your thoughts....Thanks so much 
for your feedback,Herman.