|
PureBytes Links
Trading Reference Links
|
Dave,
"... I'm partial to fully mechanical trading, if for no other reason
than because the results of an approach can be validated through
backtesting."
"Backtesting" by definition does NOT necessarily mean computer-
driven 'mechanical' methodology, although most of us ASSUME that is
what is meant!! One CAN, by hand, 'mechanically' apply rules for
generating trading signals without the aid of a computer system,
although I would readily agree that this would not be very effective,
nor efficient. I certainly would not want to labor over charts day
in and day out attempting to 'backtest by hand'!! And, 'mechanical'
does NOT necessarily mean that trading 'rules' CANNOT be 'ADAPTIVE'
to the evolutionary 'context' of time-series data. What 'mechanical'
DOES mean in the context of trading is to apply 'rules' (fixed OR
contextually-adaptive) in a DISCIPLINED manner to arrive at trading-
signal decision points. E.g., a dual-moving-average crossover system
uses the 'discipline' of 'fixed' rules to generate signals, whereas
E.W.-based systems need to use a discipline of 'contextually-
adaptive' rules. Contextually-adaptive rules CAN be stated and
specified using math and logic formulations, and to the extent that a
human mind is 'clever enough' to translate visual observation into
these math/logic formulations, that is the extent to which 'valid'
computer-driven backtesting can be accomplished.
I totally agree with you that it is highly desirable, necessary even,
to VALIDATE a trading strategy against the time-series data to which
it is applied, for the 'health' of one's account, ASSUMING the future
enfolds and evolves in a similar manner to the past. After all, why
bother to backtest if we cannot make THAT assumption at some level of
STATISTICAL SIGNIFICANCE. Attempting to trade a TOTALLY 'random'
time-series would be a rather futile endeavor, as NO trading
methodology would be very effective. So, yes, it is highly desirable
to be able to backtest a 'non-moving-target' type of strategy.
However, even THIS is a 'reasonable illusion' at best, since none of
us have crystal balls for peering into the future. It's ALL a
probability game, and the goal is to create and trade strategies and
systems whose probabilities INTO THE FUTURE may give us an 'edge'.
It's the 'reasonable illusion' that gives each of us 'faith' to
actually trade SOME kind of system. That FAITH can be derived in
many ways, including 'belief' in the statistical results produced by
a mechanical-backtesting procedure.
I'll readily admit that I am NOT, by ANY stretch of the imagination,
an 'expert' in E.W. trading. At best, currently, I'm an 'informed'
neophyte. As a matter of fact, until a couple days ago, I have
TOTALLY ignored and discounted E.W. approaches, considering them to
be 'voodoo' methodologies. But now, I DO see the potential value of
E.W. for trading, and have recently come to realize that the rules
for trading are 'context-adaptive', and MAY possibly be math/logic
encodable, even though most people see the E.W. 'principle' as a
highly 'discretionary' endeavor. So, I'm willing to re-examine my
view on this, and see how viable E.W. may be for certain kinds of
trading. From what I've learned recently, like ANY OTHER approach,
it's a PROBABILISTIC type of methodology that looks like it can 'bear
significant fruit' when one takes the time to PROPERLY navigate with
it. Would I recommend it to you? Not necessarily. Is it worth MY
time and effort to explore it? I definitely think so. Am I
an 'advocate' for E.W. trading approaches? I don't yet know, one way
or the other --- we'll see. Do I want to use AmiBroker as a tool to
find out? Absolutely, and without reservation!!!!! Hence, my post
to this forum to those who are FAR MORE skilled in AB's AFL
programming and use than am I. Will AB help me to discover what I
need to learn in a timely fashion? Absolutely, YES!! Do I need
to 'mechanically backtest' what I'm seeking to do? NO!!
So, please, let's not get into the merits of E.W. as a viable TRADING
methodology or not, or whether 'dynamically-adaptive'-ruled systems
are worthy of pursuit or not. That's a waste of this forum's
expertise, although all the dialogues HAVE been interesting,
nonetheless. I simply wanted some direction to help me determine
if 'the wheel had been already invented' using AFL ..OR.. if there
are some coding suggestions, OR cautions, when using AFL for
programming AB to do the specific job of E.W.-candidate PRELIMINARY
SCREENING --- that's all. In this regard, Dimitris's commentary has
been MOST helpful and useful, and I see I have to do more 'homework'
regarding defining the process and then implementing the
coding.
Buzz
-------------
--- In amibroker@xxxxxxxxxxxxxxx, "Dave Merrill" <dmerrill@xxxx>
wrote:
> > A complaint that I've heard many times about Advanced Get, which
does
> > a marvelous job of identifying and labeling the E.W. pattern, is
that
> > it changes the E.W.-labeled points as new data is added. Anyone
who
> > truly understands the nature of the E.W. pattern realizes that the
> > pattern DOES evolve as new information (data) is added in time.
So
> > that particular software algorithm simply reflects this 'updating'
> > process, and RIGHTLY so! One just needs to understand that this
is
> > the nature of the E.W. model as applied to financial markets.
> > Whether or not USING this DYNAMIC information for profitable
trading
> > purposes is 'practical' or not, as Dimitris has astutely pointed
out,
> > is a whole different story, and I wholeheartedly agree with this!
> > THAT issue is best left to another trading systems 'strategy'
forum.
> >
> > Buzz
>
> While others, yourself included, may use mechanical scans only to
find
> things to chart or otherwise examine manually, I'm partial to fully
> mechanical trading, if for no other reason than because the results
of an
> approach can be validated through backtesting. When signals change
as more
> data is added, this makes it very hard to know how the system would
have
> behaved at the time.
>
> Check out almost any Zig-based system -- it'll knock our socks off
with past
> data, because the signals you see from here have already
been "adjusted" by
> data that at the time hadn't happened yet. I've written a confirmed-
zig
> system, that trades only when the zig trend has progressed far
enough that
> the corner being traded won't disappear, and trust me, it's way less
> impressive.
>
> For me personally to have "faith" in an EW system, I'd want to
backtest it
> somehow. This means describing it so the actual trading signals
don't change
> over time. If they have to be delayed until they're certain, well,
that's a
> system limitation; hope it still performs well. If the signals
can't be
> stated in a stable way, how would you know if the system works well?
>
> Dave
------------------------ Yahoo! Groups Sponsor --------------------~-->
Yahoo! Domains - Claim yours for only $14.70
http://us.click.yahoo.com/Z1wmxD/DREIAA/yQLSAA/GHeqlB/TM
--------------------------------------------------------------------~->
Check AmiBroker web page at:
http://www.amibroker.com/
Check group FAQ at: http://groups.yahoo.com/group/amibroker/files/groupfaq.html
Yahoo! Groups Links
<*> To visit your group on the web, go to:
http://groups.yahoo.com/group/amibroker/
<*> To unsubscribe from this group, send an email to:
amibroker-unsubscribe@xxxxxxxxxxxxxxx
<*> Your use of Yahoo! Groups is subject to:
http://docs.yahoo.com/info/terms/
|