Testing the theory is of course fine ... However I see no degree of 
robustness in what you describe ... Backtests can be made to show 
anything one wants them to show i.e. imho backtest performance 
metrics only provide some indication of whether or not a particular 
methodology should be more thoroughly investigated.

As far as having the cability to create watchlists of a group of 
stocks meeting some criteria in this case being the best performers 
utilizing some system ... This can be done a variety of ways. 

One would assume the methodology you are describing requires that 
every so often this sort of ranking process needs to be performed 
again whether that's every day, every five years or at some frequency 
in between and that during the investigation phase you'd want to roll 
this forward from some point in time so that you can really see if 
the idea is robust or not in good markets and bad.

As a result it would seem that you need a controller that 
- Changes the AA dates and possibly other AA settings
- Runs a backtest
- Sorts the results
- Exports the results
- Reads the results storing the symbols for the top candidates in a 
watchlist
- Changes the AA dates forward
- Backtests the system on the created Watchlist with possibly other 
settings changed and/or positionscoring logic etc. i.e. a potentially 
different or at least modified version of the original AFL used in 
the original backtest.
- Performs the set of instrcutions above multiple times

This would seem to warrant a VB or J script to perform the required 
functionality and while I tend to think of BatMan ( Which is in 
essence a VBScript ) as a tool to be used for production systems / 
scans / explores it does have the capability to perform all of the 
above functions simply by filling in the blanks on a form indicating 
a job within a batch ( in the above case two jobs, one for the 
original backtest which creates the watchlist from the sorted results 
and a second to run the out of sample test using the created 
watchlist ) ... So feel free to use it or write your own flavor.

--- In [email protected], "brpnw1" <[EMAIL PROTECTED]> wrote:
>
> Hi Fred,
> 
> I'll try to clarify some more... I have a friend who claims he has 
> had great success by making his trading system do two things: 1) 
> backtest his system against recent data (he won't say how recent) 
> for a select group of stocks, and 2) create a watchlist of the best 
> performing stocks (based on the recent historical data test in #1) 
> and then the system ultimately selects from the new watchlist which 
> stocks it will trade.
> 
> I was hoping to add this type of functionality to my trading 
system. 
> I am also hoping that in the midst of all the other scripts I am 
> running for my trading system, I can somehow add a backtest script 
> that will create the additional watchlist of best-performing 
stocks. 
> 
> Whether or not my friend is correct in assuming the success of his 
> system lies in this functionality, I cannot say. I would, however, 
> like to test his theory on my own trading system. His trading 
system 
> does not use AB whatsoever. He works for a large financial firm.
> 
> Thanks,
> 
> Brian
> 
> --- In [email protected], "Fred" <ftonetti@> wrote:
> >
> > Are you referring to in backtest ?
> > 
> > --- In [email protected], "brpnw1" <tradermail@> wrote:
> > >
> > > Hi Fred,
> > > 
> > > Robust = it generates consistently high profit. Not sure if you 
> > > thought of a different definition, but this is what I meant 
> > > by "robust."
> > > 
> > > ~Bman
> > > 
> > > --- In [email protected], "Fred" <ftonetti@> wrote:
> > > >
> > > > Robust ? ... What makes it robust ?
> > > > 
> > > > --- In [email protected], "brpnw1" <tradermail@> 
wrote:
> > > > >
> > > > > REPOST (see below).
> > > > > 
> > > > > --- In [email protected], "brpnw1" <tradermail@> 
> wrote:
> > > > > >
> > > > > > Hello fellow AB crew,
> > > > > > 
> > > > > > I am looking for either search terms I can use to find 
> > > previous 
> > > > > posts 
> > > > > > on this topic (please advise which terms to use) or an 
> > > overview 
> > > > of 
> > > > > how 
> > > > > > to build the following in AFL...
> > > > > > 
> > > > > > I want to identify which specific stocks/ticker symbols 
> are 
> > > > > expected 
> > > > > > to perform best with my trading system. I would like to 
> > > isolate 
> > > > > the 
> > > > > > best-performing 20 stocks that have historically 
generated 
> > the 
> > > > > best 
> > > > > > reults for my trading system. I only want to trade these 
> > > stocks 
> > > > > with 
> > > > > > the given system.
> > > > > > 
> > > > > > My understanding is that stocks that have performed best 
> in 
> > > the 
> > > > > very 
> > > > > > recent past have a better chance of performing well in 
the 
> > > near-
> > > > > term. 
> > > > > > This is based on a friend's robust trading system which 
is 
> > > > working 
> > > > > > quite well. Therefore, I am interested in also selecting 
> the 
> > > > > length of 
> > > > > > historical data that will be tested against the system 
for 
> > > each 
> > > > > stock.
> > > > > > 
> > > > > > Any ideas or code that you can throw at me to get me 
> started?
> > > > > > 
> > > > > > Thanks in advance,
> > > > > > 
> > > > > > Brian, aka "Bman"
> > > > > >
> > > > >
> > > >
> > >
> >
>







Please note that this group is for discussion between users only.

To get support from AmiBroker please send an e-mail directly to 
SUPPORT {at} amibroker.com

For other support material please check also:
http://www.amibroker.com/support.html

 
Yahoo! Groups Links

<*> To visit your group on the web, go to:
    http://groups.yahoo.com/group/amibroker/

<*> Your email settings:
    Individual Email | Traditional

<*> To change settings online go to:
    http://groups.yahoo.com/group/amibroker/join
    (Yahoo! ID required)

<*> To change settings via email:
    mailto:[EMAIL PROTECTED] 
    mailto:[EMAIL PROTECTED]

<*> To unsubscribe from this group, send an email to:
    [EMAIL PROTECTED]

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



Reply via email to