|
PureBytes Links
Trading Reference Links
|
On Tue, 27 Jul 1999 11:57:22 -0700, you wrote:
> In fact, I suspect that by including money management
>parameters (like optimal f) in system design and testing, one will find that
>many mediocre systems we have long since dismissed because of poor win/loss
>ratios (or whatever other inadequate yardsticks we currently use to measure
>a system), would actually be ideal candidates for trading.
>
>Including money management in system testing is certainly beyond MetaStock's
>capabilities and beyond my present skills. Do you have any suggestions?
I used MetaStock for some time until I was tired of many shortcomings
showing up. So I moved back to Excel for all my calculations (trading
model (TM) set-up, goal function (GF) set-up, "genetic" optimization
(GO) routine, graphic representation of results, etc.)
At this time, I have worked out some extended GF capabilities. Based
on an article from ... (sorry, I have no access to this info at the
moment, but I'm ready to send a 600 kB zip-File of the WEB-article
(pdf) and the accompanying *.xls, if anybody is interested), I have
set-up separate Weibull distributions for wins and losses (reflecting
nicely the unsymmetrical features of these distributions). From this,
I can easily calculate e.g. the probability-to-win/loose-more-than-xx%
(p2w/lmtxx%) from past trading results. And including these into the
GF (e.g. GF = profit / p2lmtxx% * p2wmtyy%) did exactly what you
describe above: it changed dramatically my view at the trading system
optimization results.
One of my next steps will be to include the "f" into the TM for
reasons discussed above.
mfg rudolf stricker
| Disclaimer: The views of this user are strictly his own.
|