|
PureBytes Links
Trading Reference Links
|
Dear List,
I would like to ask what I hope are fair questions regarding genetic programs and neurofuzzy systems. If I am too simplistic in my understanding, I hope you will take the time to help me understand.
Here goes.. Are these types of programs super optimizated systems or distinct systems? Aren't these just brute optimizations, how do they have an advantage going forward? Don't genetic programs and neurofuzzy systems have a danger of curve fitting also? In other words if they don't have predictive features then aren't they just really good at curve fitting?
Thanks for allowing me to ask honest questions?
Dave Pyle
--- On Sun, 8/23/09, Pierre Orphelin <pierre.orphelin@xxxxxxx> wrote:
> From: Pierre Orphelin <pierre.orphelin@xxxxxxx>
> Subject: RE: Switch between systems : Yes, we can !
> To: "'David Pyle'" <dpevergreen@xxxxxxxxx>, omega-list@xxxxxxxxxx
> Date: Sunday, August 23, 2009, 6:33 AM
> Hi,
>
> Yes,you are absolutely right.
>
> The software runs up to 1,000 neurofuzzy
> systems in real time ( each of
> them having thousands of rules).
> To be more precise, we run up to 1000
> equity curves in real time, any
> equity curve could be obtained by same or
> different systems applied to
> the same or different instruments, same or
> different timeframe ( any
> combination of these in fact)
>
> These candidates will allow to
> build a dynamic managed portfolio of a
> smaller size ( or different sub portfolios).
>
> These candidates are then chosen by a
> money management scheme based on
> the permanent sorting of the candidate
> results.
> The best one is chosen every time that a
> trade close in the managed
> portfolio.
> So, the composition of the managed
> portfolio may evolve for every new
> trade ( it will depend on the performance
> of the selected and
> unselected systems over time).
>
> Because the perfect system cannot exist,
> the swapping between candidates
> seldom bring the same
> system/instrument/timeframe twice or more in a row.
>
> Until now, it's the best solution that I
> have found for performance and
> stability.
> The results are naturally improved
> by the multiple line effect ( a 10
> line portfolio will show a lowed
> MIDD than 10 contracts on one
> similar system), and the permanent
> swapping is a fairly good
> insurance against failing systems as well
> as a valid improvement of
> the multiple line smoothing effect.
>
> Depending on the composition
> of the candidate pool, the dynamic
> management scheme, and the portfolio size
> one may easily expect results
> al least similar to a big portfolio (100
> lines or more) without having
> to invest in 100 lines.
>
> The other factor is the decorrelation
> of the unmanaged equity curves.
> The higher decorrelation will produce
> the best improvements.
>
> Sincerely,
>
> Pierre Orphelin
> www.sirtrade.com
>
>
>
>
>
> -----Message d'origine-----
> De : David Pyle [mailto:dpevergreen@xxxxxxxxx]
>
> Envoyé : dimanche 23 août 2009 08:07
> À : Pierre Orphelin; omega-list@xxxxxxxxxx
> Objet : [english 100%] RE: : Switch between systems : Yes,
> we can !
>
> Pierre,
>
> Are you saying that you software runs 100 or more systems
> at a time in real
> time and switches between them in real time?
>
> Dave Pyle
>
> --- On Sat, 8/22/09, Pierre Orphelin <pierre.orphelin@xxxxxxx>
> wrote:
>
> > From: Pierre Orphelin <pierre.orphelin@xxxxxxx>
> > Subject: RE: : Switch between systems : Yes,
> we can !
> > To: omega-list@xxxxxxxxxx
> > Date: Saturday, August 22, 2009, 5:02 PM
> > Hi,
> >
> > It seems to me that you are
> > starting to discover
> > fuzzy logic with
> > you knob example.
> > You just need to implement a
> > valid solution ( the answer is not with
> > two systems, by far).
> > Well, this is a well known path to
> > us for 15 years.
> >
> > FYI the latest version of the
> > Safir-Xp2 software does dynamic
> > switching of neurofuzzy systems in real
> > time.
> > Up to 1000 trading systems
> > monitored real time that
> > would be swapped
> > at any time to build a managed portfolio
> > among these 1000 candidates (
> > usually 100 are enough).
> > Allows daily intraday range
> > bars swapping at the same time.
> >
> > More to see on the new
> > temporary pages of the web site:
> > www.sirtrade.com
> >
> > There are several animated tutorials
> > that explain the new thing.
> >
> > Does it work ?
> > See by yourself...Zero programming
> > needed.
> > Even Not a single line of EL
> > to write from the systems
> > development
> > (done while you sleep) to to automated orders.
> > And of course, it's back testable.
> >
> > We will publish real time results
> > within a few weeks.
> > I know, I said the same 4
> > years ago...
> > But we needed more than
> > 4 years to complete this project.
> >
> > Just let we know if you already had
> > seen something similar.
> > I bet you will not...
> >
> >
> >
> > Sincerely,
> >
> > Pierre Orphelin.
> >
> >
> > -----Message d'origine-----
> > De : Mark Johnson [mailto:janitor@xxxxxxxxxxxx]
> >
> >
> >
> > Here's a fourth suggestion: replace the word "switch"
> > by the word "knob". Instead of restricting yourself
> > to only two possibilities:
> > (100% of system A, 0% of system B)
> > (0% of system A, 100% of system B)
> >
> > allow yourself to imagine trading X% of system A and
> > (100-X)% of system B. Imagine a "knob" that you can
> > smoothly rotate, from (A=100, B=0) to (A=0, B=100).
> >
> > Now imagine using technical analysis indicators to
> > choose an appropriate value of X%.
> >
> >
> >
> >
>
>
>
>
>
>
|