|
PureBytes Links
Trading Reference Links
|
Gosub states:
<<However.... What I'm really looking for is what people found
to be the best indicator to help identify the current situation (not
the future). I was basicaly throwing out a question to everyone to
see if anyone has had better success than me when developing a
transitional system that identifies when it is necessary to change to
a different indicator.>>
That is a fair question and I'll be happy to share my own experience
with backtesting (and trading) trend detection systems.
In the book Street Smarts co-authored by Linda Raschke, she outlined a
trading methodology named the 'Holy Grail' which uses the ADX, PDI,
MDI indicators to determine trend direction.
With the help of a friend who has more 'instinctive' abilities than
me, I developed a backtest system based upon a slight derivitive of
Linda's 'Holy Grail' and hit paydirt.
The 'Holy Grail' system is a classic pullback system that Linda
primarily trades futures with. My friend and I adapted it to stock
trading.
In my own software, I calculated an Annual Percentate Rate of return of
189.59% for the past 10 or 11 years on the whole market. I recently
adapted my 'Holy Grail' system to Amibroker and achieved silimar
results but with improvements using the Optimize function. AB
calculates the APR differently and reports a RAR of 142.21%, but with
even greater monatery return amounts. While I don't know for sure, I
think the difference in calculated percentage returns is that in my
own system calculates returns based upon calendar days of exposure to
the market, while AB might use Bars of exposure to the market. But
that is a subject for another discussion.
A few weeks ago, I believe that Jayson stated that pullback systems
stopped working after the 1999 blowoff, buy my system has shown a
profit every year for the past 11 years.
As an aside... I'd encourage others to post backtest results on the
whole market for the past 10 or 11 years of their favorite systems
just to see how good other trading systems can be using a RAR or APR
measurement.
But anyway, back to this thread:
In developing other trading systems, I can generally improve results
by incorporating the ADX, PDI, MDI indicators to implement trend
detection in my system development efforts.
Also, I use a Quotes Plus database which has a historical QRS
indicator which emulates Investors Business Daily's proprietary
Relative Strength indicator. And my experience has been that by also
filtering with the QRS indicator, my backtesting results are improved
even more. The only problem is that Quotes Plus apparently
incorporated the QRS measurement in 1995, so when using that
indicator, you cut into the amount of historical data that you can use
to backtest.
Hope this is what you were asking for. And I hope that others will
share their 'favorite' trend detection indicators with the metrics
they use to measure system trade improvements. Whenever I need a trend
indicator, the ADX, PDI, MDI are the first indicators I grab for. How
about the rest of you?
Phsst
--- In amibroker@xxxxxxxxxxxxxxx, "gosub283" <gosub283@xxxx> wrote:
> Hi Yuki,
>
> Thanks for your comments.
> I agree with you completely. However....
> What I'm really looking for is what people found
> to be the best indicator to help identify the
> current situation (not the future).
> I was basicaly throwing out a question to everyone to
> see if anyone has had better success than me when
> developing a transitional system that identifies
> when it is necessary to change to a different indicator.
> I have had success with such a system. The problems
> arise during the transition phase from Trading to Trending.
> (Some losses due to ambiguity/confusion.)
> Once the transition is completed and the system automatically
> changes to the appropriat indicator, it works fine.
> I was trying to see if anyone else had played with this
> and perhaps found a better way.
> I am familiar with many indicators which can help with this
> but I'm always open to others.
> Jayson was very helpfull in pointing out the R-Squared and
> Twiggs-Money Flow.
>
> Thanks again,
> gosub283
>
>
> --- In amibroker@xxxxxxxxxxxxxxx, Yuki Taga <yukitaga@xxxx> wrote:
> > Hi gosub283,
> >
> > Thursday, March 27, 2003, 2:08:16 AM, you wrote:
> >
> > g> When a human looks at a chart, he/she can immediately determine
> if
> > g> a market is in a TRENDING mode or a RANGING mode. It is a most
> > g> amazing feat of human visual data analysis that takes place in a
> > g> matter of seconds. Trying to get computers to "visually" analize
> > g> anything takes major computing power. Unfotunately computers and
> > g> trading system have a much more difficult time of determining
> > g> these market modes than us humans.
> >
> > I am not so sure humans do a much better job of this analysis than
> do
> > computers. When we look at charts, we are looking at 100 percent
> > history. It is indeed rather easy to "segment" this history into
> > trending and ranging periods using analytical abilities. Yes, it
> > does happen in seconds, and seems almost effortless.
> >
> > However, the most important time period for any trader is that
> period
> > of time which is NOT VISIBLE on the chart -- namely, the future. We
> > can "see" the ranging and trending periods, but only in retrospect,
> > where our minds make nice, neat visual categories for us based on
> > what we *see*. But at the "real time" turning points, when the
> > actual transition is in the future, out of sight, I doubt if either
> > human or computer can reliably identify these transitions in
> advance.
> > (Reliably to me would be with some statistical edge that clearly
> > beats random guessing.)
> >
> > In hindsight, it is all so clear of course -- so clear that it is
> > tempting to believe that one can find some mathematical formula that
> > will reveal all. But this is a sect in the cult of "Holy Grail-ism"
> > in my opinion. To test my theory that human beings are not any
> better
> > at identifying the transitions in "real time" than computers, scroll
> > a chart very slowly (bar by bar) from left to right, with the future
> > always off the screen, just as in real life. Make sure the chart is
> > a completely unknown issue or index, and preferably one with the
> > dates entirely removed or obscured (to do this test correctly
> someone
> > else will probably need to set up a blind test for you, I don't
> think
> > you can truly set up a valid blind test yourself). Now see
> how "good"
> > your human brain is at identifying the transitions in "real time".
> > And good luck! ^_-
> >
> > Yuki
------------------------ Yahoo! Groups Sponsor ---------------------~-->
Make Money Online Auctions! Make $500.00 or We Will Give You Thirty Dollars for Trying!
http://us.click.yahoo.com/yMx78A/fNtFAA/46VHAA/GHeqlB/TM
---------------------------------------------------------------------~->
Send BUG REPORTS to bugs@xxxxxxxxxxxxx
Send SUGGESTIONS to suggest@xxxxxxxxxxxxx
-----------------------------------------
Post AmiQuote-related messages ONLY to: amiquote@xxxxxxxxxxxxxxx
(Web page: http://groups.yahoo.com/group/amiquote/messages/)
--------------------------------------------
Check group FAQ at: http://groups.yahoo.com/group/amibroker/files/groupfaq.html
Your use of Yahoo! Groups is subject to http://docs.yahoo.com/info/terms/
|