Basic philosophy
explained.
There are number of well known technical
analysis software packages that can perform plenty of number
crunching on the basis of parameters defined by the user. Unfortunately,
these standard TA software packages have little or no
ability to optimize their internal parameters for
better performance of real trading. They can calculate and
display 100+ technical indicators on the basis of predefined internal
parameters. They urge user for ‘do it yourself curve fitting’ with the
recommendation to select suitable values of internal parameters. A
diligent user of such software package is trying hard to find optimal
values of parameters. He is experimenting with various values of
parameters, asking friends for ‘hot tips’, searching for ‘hot tips’ on
Internet and losing his or her leisure time on activities, which
could be done automatically with some more sophisticated software
package. The question is, why the software vendors have not defined such
optimal parameters in the software itself, or in the user manual? The
answer is very simple – they are not able to do so, because
there are no universally valid values of parameters for
calculating technical indicators!
Many vendors of technical analysis software
are trying to make an impression that their software is better because
it can calculate more technical indicators than other software packages.
The question is why they offer so many technical indicators and not only
one technical indicator, which is the best of all. The answer is also
very simple. There is no such technical indicator with universal
validity and relevant trading performance. Any technical
indicator is only an oversimplified model of stock trading with no
universal validity and with no relevant trading performance.
I notice occasionally various remarks about
‘curve fitting’. Authors of these remarks are trying to make an
impression that curve fitting is something wrong, which is artificially
improving the back testing performance of the technical analysis
software, with no impact on the performance of real trading. On the
other hand, these ‘anti curve fitting gurus" do not hesitate to search
for hot tips and rumors about formulas, suitable values of parameters
for calculating various technical indicators or use their own ‘secret
hot tips’ in hope of achieving better trading performance. These
activities in fact can also be regarded as curve fitting, namely, as ‘do
it yourself curve fitting’ on the basis of rumors. For example,
questions on number of days for calculating moving averages are very
frequent.
There is also another approach to curve
fitting, namely curve fitting on the basis of mathematical optimization.
Curve fitting on the basis of mathematical optimization is
natural and useful process of learning, during which unknown
parameters of the mathematical model of stock trading are determined in
order to achieve the best back testing performance and thus improving
the performance of the real trading. The real danger is not
in the curve fitting, but in the mathematical model of trading
itself, which in extreme negative cases can either:
- perfectly fit to past trading history,
with no ability to generate profitable trading recommendations for new
data or
- badly fit to past trading history, also
with no ability to generate profitable trading recommendations for new
data.
Theoretically perfect way of generating buy -
sell recommendations is to have a mathematical model or theory of
trading without any internal parameters, which fits to all stocks and
all situations of the stock market and requires no learning on
historical data. Of course, theoretically perfect model or
theory of trading is not available and belongs into the realm of science
fiction.
The best, really available way of generating
buy - sell recommendations is to use adaptive mathematical models of
trading with small number of internal parameters, capable to optimize
these internal parameters on daily basis, on the latest historical
data. StockMarketMirror software is based on the adaptive
mathematical models of trading with small number of internal parameters,
that are optimized daily on historical data.
The absence of parameter optimization
and absence of back testing function in the technical analysis software
package is almost 100% guarantee of bad performance of real
trading. Excellent back testing performance of any technical
analysis software is necessary precondition (but not
sufficient) for high quality technical analysis software, which is
worth trying and eventually using on the regular
basis.
Low purchasing price of technical analysis
software is another marketing factor used to attract users. Of course
the purchasing price must be affordable, but the most important factor
is the bottom line after one or more years of trading. If the software
purchase price is $30 and you lost $5000 after one year of trading, the
bottom line is minus $5030. If you pay $200 for software and make profit
of $5000 by using it with stock trading, your bottom line is plus $4800.
Some users may want to calculate their leisure time, spent by the
computer using the software package or searching for ‘hot tips’ into
their expenditures. In this case, software features like automatic (or
single click) generating of buy - sell recommendations plays important
role. StockMarketMirror software can generate all its
recommendations with a single mouse click and even automatically, in
cooperation with Task Scheduler and its price is
affordable.
One remark, frequently seen on internet is
asking, why software vendors are selling their technical analysis
software, instead of becoming multimillionaires by using their own
software. They are adding, that there is no stock technical analysis
multimillionaire and therefore, all technical analysis is a dubious
pseudo teaching. People, making those malicious remarks do not care to
name a single software author or software package that is doing such
claims of becoming multimillionaire by using their software. The
reasons, why software vendors are developing technical analysis software
are the same as in the case of other software. Professionals
are developing software in order to earn for a living by some useful
activity and to make some profit. Amateurs are developing software
because they like it or they are trying to become professionals.