|
PureBytes Links
Trading Reference Links
|
Hi Paul, maybe Roy, Preston, or someone else will have time to get rid
of a couple of prev's for you.
With prevs on my computer it takes about 1 hour and 35 minutes to
process 1500 stocks. Using a DLL, it only takes a couple of minutes.
I'm very familiar with Stridsman's work, and the systems that he has
made public. He really doesn't share much information on the systems
he actually uses. He always puts one or two so, so performers in his
books to keep readers interested since 99% of the systems he does
write about perform poorly.
One tip on applying this formula (sorry if I'm repeating myself, but
it's very important to the overall returns someone gets) is it
performs much better.
As an example, I combined the DBS entry set to st dev 30 min 20 and
max 60. I used the DBS entry only and a Hull Moving Avg exit set to
250 with a 10% stop loss. I tested 2004 as random pick for a year. As
a trading year, 2004 was that great depending on what you were trading.
On the S&P 500, the system returned 8% with 38% wins, 2.2% stopped out
and a win/loss ratio of 2.4%. Relative drawdowns were less than 1%.
On the Valueline Timeliness 100 stocks for 2004, the same system
returned 27% with 47% wins, 4.6% stopped out and a win/loss ratio of
2.9%. Relative drawdown was less than 1%.
What this says is that a winning trade compared to a losing trade from
the S&P 500 returns nearly the same amount as does the same
combination from the Value Line list, but the system finds more
winners in the Value List so the overall return is much higher. When
comparing the win/loss ratio of 2.4% to 2.9%, the VL stocks out
perform the S&P stocks but the real difference as I said was in the
number of winners.
I suspect the same results would take place if the S&P Platinum list
or IBD list were used, but I didn't test them.
Your test results might be different depending on your exits, time
frames, universe of stocks, and other stuff, but the comparison is
good for at least a relative look at the difference between using a
screened and un-screened universe of stocks.
Stridsman didn't mention this in his books.
Hope someone can fix those prevs.
Super
--- In equismetastock@xxxxxxxxxxxxxxx, pumrysh <no_reply@xxx> wrote:
>
> Paul,
>
> I believe we can indeed give you some answers.
>
> A little history behind the indicator. Originally it was part of
> some work done by Thomas Stridsman for Futures Magazine...I believe
> that was in 1998. It 2004 it appeared in Traders Mag. Here is a
> summary:
> "Dynamic Breakout System" (Trading System Lab, Feb. 2003)
> Author: Thomas Stridsman
> Summary: This system enters a long (short) trade if the last closing
> price is above (below) the highest high (lowest low) of the lookback
> period. The lookback period changes based on market volatility.
>
> The Tradestation folks have written a very good discussion(PDF) on
> the indicator which I will upload into our files section for you.
>
> Years ago, before this forum and many others were around the folks
> at Equis hosted a forum that I was simply known as Metastock. A
> fellow by the name of Walter Lake introduced Stridsman's work to the
> group back in '98. There was also a bit of discussion about the use
> of +prev-prev. Turns out that it is the key to making the
> indicator "dynamic" or what we call variable in today's terminology.
> Luckily we can still go back and review some of those earlier posts
> and the site listed below will get you there.
>
> http://www.purebytes.com/cgi-local/swish/swish-cgi.pl
>
> Search for DBS
>
> Metastock
>
> 1998
>
> This will pull up all the old discussions about the system.
>
> Also
>
> http://www.purebytes.com/archives/metastock/1998/msg06413.html
>
> From Walter:
> -----snip-------
> The "+Prev-Prev" comments in the article should be researched when
> using "DBS Lookback" in other Metastock indicators.
>
> Here's one writer's experience ... "I tested both versions on several
> indices and found them to be different too:
> This +PREV-PREV causes a shift to the left of the indicator, a trick
> to center an indicator?"
>
> "after some testing, I found that without +PREV-PREV LastValue will
> deliver a constant value (the last value for LookBack in Your chart).
>
> So in writing Tema(C,LastValue(Fml("LookBack"))) You get a TEMA with
> a constant period, as required by Metastock.
>
> By using Tema(C,LastValue(Fml("LookBack")+PREV-PREV))
> You get what You want, a TEMA with a variable period. Nice trick!"
>
> Walter
> ----snip-----
>
> Realize that back then we didn't have DLL's so our only recourse was
> to get a little creative with indicators. Today we have DLL's and it
> is for just such a purpose that they should be used.
>
> The answer then to removing the prevs is to use a DLL.
>
> Lnc:=Ref(HHV(H,LastValue(Fml("DBS-LookBack")+PREV-PREV)),-1);
>
> using the equis forum DLL available in our files, EquisForum, or
> Traders Consortium the formula would then become:
>
> LNC:=Ref(ExtFml("ForumDLL.VarHHV",H,Fml("DBS-LookBack")),-1);
>
> You will notice that LastValue is no longer used as well.
>
> That takes care of the majority of prev statements and should speed
> up your work quite a bit. If you would like I will continue to look
> at DBS-Lookback indicator and see if that prev can be removed as
> well if you like.
>
> After reading the Tradestation PDF entitled Dynamic Breakout II
> Strategy I think you will find that this system certainly has some
> merit although it does need some work.
>
> Glad we could help,
>
>
> Preston
>
>
>
> --- In equismetastock@xxxxxxxxxxxxxxx, ppappppapp <no_reply@>
> wrote:
> >
> > Greetings all,
> >
> > A 'little bird' suggested that I should post my querry here and
> > hopefully someone might offer some advice.
> >
> > O.k. here goes;
> >
> > The following indicators are based on the Dynamic Breakout system
> > (DBS) as found in Thomas Stridsman's book "Trading systems that
> work"
> >
> > It's simply a variation on the Turtle channel breakout method
> where
> > you buy a breakout above the highest high of `x' number of days
> ago.
> > (i.e. a HHV channel)
> > This DBS system varies the channel width dynamically according to
> > price volatility (as measured by standard deviation)
> >
> > The following indicator finds the VALUE of the varying channel
> width
> > (in this case the HHV channel will vary between 20 and 100 bars
> ago)
> > and is then referenced into the second indicator
> >
> > ******************************
> >
> > {DBS-Lookback}
> >
> > {-----User Inputs-----}
> >
> > VolPer:=Input("Enter StdDev periods",1,100,30);
> > LBmin:=Input("Enter Min. lookback periods for entry",5,100,20);
> > LBmax:=Input("Enter Max. lookback periods for entry",20,200,100);
> >
> >
> > HV:= Stdev(C, VolPer);
> > YestHV := Ref(HV, -1);
> > DeltaHV := 1+((HV-YestHV)/HV);
> > If(Cum(1)=1,LBmin,Min(Max(PREV*DeltaHV,LBmin),LBmax));
> >
> > ********************************
> >
> > This following second indicator then references the above first
> > indicator to plot entry and exit signals
> > (i.e. not `clean' signals but ALL valid signals where they occur)
> >
> > ********************************
> >
> > {DBS-Lsigs}
> >
> > {Dynamic Breakout system - Long signals}
> >
> > { User inputs }
> > Mess1:=Input("Change defaults in DBS-Lookback Indicator",0,0,0);
> >
> > {Long entry & Long exit Channel}
> > Lnc:=Ref(HHV(H,LastValue(Fml("DBS-LookBack")+PREV-PREV)),-1);
> >
> > Lxc:=Ref(LLV(L,LastValue(Fml("DBS-LookBack")+PREV-PREV)),-1);
> >
> > { Entry/Exit Long breakouts }
> > entry:=H>Lnc;
> > exit:= L<Lxc;
> >
> > entry;
> > -exit
> >
> > *******************************
> >
> > o.k., now for the question;
> >
> > You will note from the above, the use of one(1) PREV command in
> the
> > first indicator and four(4) PREV commands in the second indicator
> (as
> > well as the PREV command associated with the referenced formula
> > within this indicator)
> >
> > Although the indicators work fine and dandy, they take a long,
> Long,
> > L?--O--?N?--G time to plot on screen. Obviously due to the large
> > amount of computer resource demanded by the PREV command.
> > If I attempt to run an exploration on numerous stocks/futures
> using
> > these indicators, I generally have to do it at night just before
> bed.
> > The results are available for me in the morning. (lucky I sleep
> for
> > 8+ hours)
> >
> > Is there anything I can do to the indicator code to speed up the
> > execution?
> > Or is this simply a Metastock limitation that I will have to live
> > with?
> > In short how can i replace the PREV commands?
> >
> > Any tips would be much appreciated.
> >
> >
> > Thanks kindly,
> >
> > Paul
> >
>
Yahoo! Groups Links
<*> To visit your group on the web, go to:
http://groups.yahoo.com/group/equismetastock/
<*> Your email settings:
Individual Email | Traditional
<*> To change settings online go to:
http://groups.yahoo.com/group/equismetastock/join
(Yahoo! ID required)
<*> To change settings via email:
mailto:equismetastock-digest@xxxxxxxxxxxxxxx
mailto:equismetastock-fullfeatured@xxxxxxxxxxxxxxx
<*> To unsubscribe from this group, send an email to:
equismetastock-unsubscribe@xxxxxxxxxxxxxxx
<*> Your use of Yahoo! Groups is subject to:
http://docs.yahoo.com/info/terms/
|