Forum Archive Index - June 2001

[Phaedrus <Phaedrus@techemail.com>]

John,
      It's Henley charts. I can tell. They can sue me.
 The basic problem here is that you have not understood the nature of 
Optimisation, its advantages, disadvantages, and pitfalls. If you have this 
option turned on and left on, the program will always be trying to find the 
most advantageous moving average values to use. Price movements can and do lead 
to different values being applied as time goes by. Every time the system 
changes the values being used, all signals generated by the old values are 
scrubbed, and an entirely new set applied. The quick fix to stop this 
disconcerting event happening is as follows:- Run the optimisation program, and 
note the moving average values it comes up with. Enter these values as the 
manual defaults, and always use those. All signals generated will be permanent 
and unchanging, no matter what.  With a good software package, you can optimise 
for the best Hit rate, or the greatest % gain, or the best Win/Loss ratio, or 
the lowest drawdown figure,  or the lowest number of consecutive losses, etc. 
You have turned yourself and your fortunes over to an ignorant machine running 
a very simple system - how can a single set of parameters be appropriate for 
all traders? Say both you and I wanted to trade GPG using a double ma crossover 
system. (I would never use this, or any single indicator in isolation, but we 
will ignore that for now.) You, I think, would want a very reliable system, 
with a high hit-rate, and a small number of wrong signals, and not too many 
signals (not too active). This would give a system with a lower total return 
than one I would want, which would have many signals, many of them wrong, but 
every trend change would be caught very early and exited early - prematurely in 
some cases, but overall, this should give a higher overall return than yours. 
Clearly we cannot both use the same settings.
 "Do all of these programs change the results as more information is gained 
over time?" No, unless you are continually re-optimising the parameters, in 
which case, Yes.
 "The idea of buying and selling on cross overs is great in theory but is it 
always true?" No. Nothing is always true. There will always be a percentage of 
false signals.
  I tested a double moving average crossover system on GPG on my database, 
optimising for maximum gain. The results (using simple moving averages) were as 
follows :- optimum ma values 7 and 67. Average gain 13% pa. (Buy/Hold gain was 
9% pa) 41% of trades were profitable, 59% of trades resulted in a loss. The 
maximum number of consecutive losing trades was 4. Using Exponential moving 
averages, the results were ma 10/62, 8% pa, other results same as simple ma. 
Now, you tell me, - is this moving average system worth using on this stock?
  My advice would be to examine the results of your backtesting more 
critically. If you can't beat buying and holding by a significant margin, why 
bother trading? It is very risky to use a single indicator in isolation - don't 
do it. Get confirmation of signals from other indicators, preferably of a 
different class. An example would be a trend indicator, an oscillator and a 
volume indicator. Read more books. Get better software.
                                      Phaedrus.

_____________________________________________________________
Are you a Techie? Get Your Free Tech Email Address Now! Visit 
http://www.TechEmail.com


----------------------------------------------------------------------------
http://www.sharechat.co.nz/          New Zealand's home for market investors
----------------------------------------------------------------------------
To remove yourself from this list, please use the form at
http://www.sharechat.co.nz/forum.shtml.