[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[amibroker] Re: Real World Systems - Multiple Sub Signals



PureBytes Links

Trading Reference Links

I don't think this is a math question per se.  If you are using 
multiple indicators as a voting system or as a scaling up/down system 
and those indicators do NOT have a high degree of correlation then 
this is probably a good thing, if they do it probably isn't.

--- In amibroker@xxxxxxxxxxxxxxx, "Ken Close" <closeks@xxxx> wrote:
> Fred: thanks, and I realize what you say is true.  My question was a
> mathematics one---is it reasonable to expect dd to decrease when 
added
> components are brought into a combination signal?
> 
> Seems like something that says that stnd dev of averages are less 
than
> the stnd dev of the components of the averages is at work....
> 
> Ken
> 
> -----Original Message-----
> From: Fred [mailto:fctonetti@x...] 
> Sent: Sunday, April 13, 2003 5:11 PM
> To: amibroker@xxxxxxxxxxxxxxx
> Subject: [amibroker] Re: Real World Systems - Multiple Sub Signals
> 
> Ken,
> 
> Multiple sub-signals as it were have long been around and are 
> typified by Ultra Systems software.  If you think about it even 
> simple systems like RUTTR use this principle in that the MACD & 
> Stochastic must agree.  This concept can be used a lot of different 
> ways.  It can be used as a voting system like RUTTR does where 
there 
> must be a unanimous vote before taking action or it can be used to 
> decide to what level one should be invested i.e. 25%, 50% 100% etc.
> 
> --- In amibroker@xxxxxxxxxxxxxxx, "Ken Close" <closeks@xxxx> wrote:
> >  Multiple Sub Signals: - here is a real world trading approach 
that 
> I
> > have never seen mentioned here.  I would like to get some 
reaction.
> > This is based on some real world trading that is happening on 
> another
> > platform that I am helping port over to AB.
> > 
> >  
> > 
> > The idea is to have signals for multiple subsystems and then take 
> your
> > trading signal when a majority of the subsignals "line up".
> > 
> >  
> > 
> > For discussion purposes, visualize a mov avg cross over   AND
> > 
> > A volume oscillator   AND
> > 
> > An advance decline curve   AND
> > 
> > Perhaps a VIX type signal.
> > 
> >  
> > 
> > If you let each one be a buy or sell, and call each S1, S2, S3, S4
> > 
> >  
> > 
> > then your buy statement could be 
> > 
> >  
> > 
> > Buy = S1 AND S2 AND S3 AND S4.
> > 
> >  
> > 
> > This might be a little too stringent, so perhaps you code it to 
> give a
> > buy if 3 of the 4 signals are a buy and you do not care which 
ones.
> > 
> >  
> > 
> > In the work that I am doing with this approach, I am seeing that 
> each
> > S(i) has a return and a dd over a long period of time, but as you 
> add
> > combinations of the signals, the return increases a little bit 
but 
> the
> > dd seems to drop and drop quite a bit.  An example might be 
returns 
> for
> > each one individually of say 8-10% CAR and 13-15% dd, but when 3 
> out of
> > 4 are combined as I describe above, the resulting return might be 
9-
> 12%
> > CAR and 6-9% dd.   These are not barn burners, and those seeking 
or
> > actually trading 50-100% CAR systems will laugh as they hit 
delete, 
> but
> > for some folks who are risk adverse, or managing retirement 
> portfolios,
> > or whatever, this kind of approach might have some appeal.
> > 
> >  
> > 
> > In the work I have done so far, there has been NO OPTIMIZATION of 
> the
> > parameters within the subsystems.  Any In sample period compares
> > favorably with OOS periods.  The results look steady over 12 
years 
> of
> > data, with variations due to changing market conditions.  (I am 
not 
> sure
> > "IS and OOS" apply when no optimization has been done, but what I 
> mean
> > is that breaking the total time period up into sections shows no 
> real
> > degradation or "blowup" of one period relative to the other.)
> > 
> >  
> > 
> > The details of the subsystems are not the issue here-what I am 
> raising
> > is the question of the drawdown dampening effect of the 
> combination.  Is
> > this a mathematically correct thing to expect?  Does moving in 
this
> > direction promise dd reduction?   Assuming you select subsystems 
> that
> > are not highly correlated, should you not expect improvements in 
the
> > combination that are not possible in the single subsystems?
> > 
> >  
> > 
> > Any comments?  Build-upons?  What?
> > 
> >  
> > 
> > Ken
> 
> 
> 
> Send BUG REPORTS to bugs@xxxx
> Send SUGGESTIONS to suggest@xxxx
> -----------------------------------------
> Post AmiQuote-related messages ONLY to: amiquote@xxxxxxxxxxxxxxx 
> (Web page: http://groups.yahoo.com/group/amiquote/messages/)
> --------------------------------------------
> Check group FAQ at:
> http://groups.yahoo.com/group/amibroker/files/groupfaq.html 
> 
> Your use of Yahoo! Groups is subject to
> http://docs.yahoo.com/info/terms/


------------------------ Yahoo! Groups Sponsor ---------------------~-->
Make Money Online Auctions! Make $500.00 or We Will Give You Thirty Dollars for Trying!
http://us.click.yahoo.com/yMx78A/fNtFAA/i5gGAA/GHeqlB/TM
---------------------------------------------------------------------~->

Send BUG REPORTS to bugs@xxxxxxxxxxxxx
Send SUGGESTIONS to suggest@xxxxxxxxxxxxx
-----------------------------------------
Post AmiQuote-related messages ONLY to: amiquote@xxxxxxxxxxxxxxx 
(Web page: http://groups.yahoo.com/group/amiquote/messages/)
--------------------------------------------
Check group FAQ at: http://groups.yahoo.com/group/amibroker/files/groupfaq.html 

Your use of Yahoo! Groups is subject to http://docs.yahoo.com/info/terms/