|
PureBytes Links
Trading Reference Links
|
<span lang=EN-GB
style='font-size:11.0pt;font-family:Arial;mso-ansi-language:EN-GB'>Guys <span
class=SpellE>guys guys...this is getting a little
bit out of hand and is getting very confusing...so I will start <span
class=SpellE>again..and BTW I did not make any
contribution concerning the SR zig <span
class=SpellE>zag. Spyros is the one who wrote
that indicator. Even I don't have that indicator.
<span lang=EN-GB
style='font-size:11.0pt;font-family:Arial;mso-ansi-language:EN-GB'>So I start
again:
<span lang=EN-GB
style='font-size:11.0pt;font-family:Arial;mso-ansi-language:EN-GB'>I'm made a
little formula with the peak and trough function.
<span lang=EN-GB
style='font-size:11.0pt;font-family:Arial;mso-ansi-language:EN-GB'>If I <span
class=SpellE>backtest on historical data from intraday to daily on any
stock commodity or index it is showing very big profits (1 minute 5 minutes 1
hours does not matter) and every time with the same parameters (almost the holy
grail).
<span lang=EN-GB
style='font-size:11.0pt;font-family:Arial;mso-ansi-language:EN-GB'>BUT...when I
start to use this formula in the advisor and in real time, now and then it
gives a signal that 2 or 3 or 4 bars later disappears like it never took this position.
Also the tester confirms this. When the advisor gives a signal and I run immediately
my tester he also show this signal...when the advisor takes the signal away and
I run my tester again it also pretends it never took this position. No need to
tell you that the signals that are taken away (pretending they never existed)
are loosing trades.
<span lang=EN-GB
style='font-size:11.0pt;font-family:Arial;mso-ansi-language:EN-GB'>A mail that I
found of Owen Davies ( at the bottom of this mail) is
explaining a similar problem. On the mail of Owen there was only 1 answer from
Al Taglavore who said that this problem is caused due
to the zig zag indicator
witch is used in the peak and trough function. I know that also but is there a
way to overcome this problem. A suggestion that I made is maybe the SR <span
class=SpellE>zig zag (confirms that the <span
class=SpellE>zig zag is giving a 100% reliable
signal) but I don't have this SR zig <span
class=SpellE>zag function. AND...maybe there is
someone who knows an other solution.
<span lang=EN-GB
style='font-size:11.0pt;font-family:Arial;mso-ansi-language:EN-GB'>Off course
with a little bit of conmen sense I can already predict once you place this SR <span
class=SpellE>zig zag function in the peak and
trough formula the golden mountains will disappear and make place for little
sand dunes. <span style='font-weight:
bold;mso-bidi-font-weight:normal'>But still I would like to know when a peak or
trough signal can be considered as 100% <b style='mso-bidi-font-weight:
normal'>true.
<span
style='mso-spacerun:yes'>
<span lang=EN-GB
style='font-size:11.0pt;font-family:Arial;mso-ansi-language:EN-GB'>greetings<font
face=Arial> <span
style='mso-spacerun:yes'> Igor
<span lang=EN-GB
style='font-size:10.0pt;font-family:Arial;mso-ansi-language:EN-GB'>
<span lang=EN-GB
style='font-size:10.0pt;mso-ansi-language:EN-GB'>
<span lang=EN-GB
style='font-size:10.0pt;mso-ansi-language:EN-GB'>To: metastock@xxxxxxxxxxxxx>
Subject: Peak and
<span style='font-size:
10.0pt'>> trough> Date: Wednesday, October 24,
2001 2:58 PM> > Among the many
<span style='font-size:
10.0pt'>> things I don't understand, this one has> been bothering me of
late:> > A
<span style='font-size:
10.0pt'>> while back, I decided to check one of my assumptions> and test
the
<span style='font-size:
10.0pt'>> higher-high, higher-low/lower-high, lower-low> definition of
trends.
<span style='font-size:
10.0pt'>> The easy way was to create a system> using peak() and
trough(). It
<span style='font-size:
10.0pt'>> worked beautifully.
Virtually> any contract I ran the system past, it
<span style='font-size:
10.0pt'>> made money. This> I
took to confirm the validity of the trend
<span style='font-size:
10.0pt'>> definition.> > Then the obvious dawned on me:<span
style='mso-spacerun:yes'> Why not see whether
<span style='font-size:
10.0pt'>> there> was enough of the move left, on average, to make a buck
from it>
<span style='font-size:
10.0pt'>> after the peak or trough was far enough behind us to get the>
signal in
<span style='font-size:
10.0pt'>> real time? I wrote
another system that included a delay> factor, so
<span style='font-size:
10.0pt'>> that one would enter or exit a trade only when the> price had
retraced
<span style='font-size:
10.0pt'>> from the peak or trough by the appropriate> percentage.<span
style='mso-spacerun:yes'> Again, it
<span style='font-size:
10.0pt'>> worked just fine. In
historical testing, it> made money like magic on
<span style='font-size:
10.0pt'>> anything from 5-minute to daily bars.> > Problem:<span
style='mso-spacerun:yes'> When I put it on
<span style='font-size:
10.0pt'>> real-time data, it gave a lot of bad> signals.<span
style='mso-spacerun:yes'> Then it suddenly
<span style='font-size:
10.0pt'>> recalculated things, decided that the> minor up and down trends
of the
<span style='font-size:
10.0pt'>> last few weeks--this was> on smallish intraday bars--had really
been a
<span style='font-size:
10.0pt'>> long up trend,> gave a new set of signals, and declared itself
a
<span style='font-size:
10.0pt'>> winner.> > Does anyone understand these functions well
enough to>
<span style='font-size:
10.0pt'>> explain this behavior to me?
I knew that peak() and trough()> backdate
<span style='font-size:
10.0pt'>> their results by putting their signal several bars> before it
was
<span style='font-size:
10.0pt'>> possible to receive it; that is what I was trying> to correct
with the
<span style='font-size:
10.0pt'>> delay factor. Now it
seems that they> also recalculate their old
<span style='font-size:
10.0pt'>> percentages by comparing against> the latest data rather than
limiting
<span style='font-size:
10.0pt'>> themselves to the data> that available in real time.> >
No doubt this is
<span style='font-size:
10.0pt'>> a real beginner's mistake (despite having> played with this for
years),
<span style='font-size:
10.0pt'>> but it would have seemed> reasonable to assume that a change of
X% three
<span style='font-size:
10.0pt'>> weeks ago> should remain X%, even if we looked at it
later. This sort>
<span style='font-size:
10.0pt'>> of thing has to be seen within its context, or it's useless.>
Is there
<span style='font-size:
10.0pt'>> some reason the functions have to be written this way,> which
I'm
<span style='font-size:
10.0pt'>> completely overlooking, or did someone just> butcher this piece
of
<span style='font-size:
10.0pt'>> code?> > Many thanks.> > Owen Davies
<span style='font-size:
10.0pt'>> : 19/04/02
<span
style='font-size:10.0pt;color:black'>
<span
lang=EN-GB style='font-size:10.0pt;color:black;mso-ansi-language:EN-GB'>Perhaps
the problem can be
<span
lang=EN-GB style='font-size:10.0pt;color:black;mso-ansi-language:EN-GB'>>
assigned to the Zig Zag
feature that is used inPeaks and Troughs.<span
style='mso-spacerun:yes'> Refer
<span
lang=EN-GB style='font-size:10.0pt;color:black;mso-ansi-language:EN-GB'>> <span
class=GramE>to MetaStock manual (<span
class=SpellE>ver 7.03) page 528:"Be forewarned, that the last
<span
lang=EN-GB style='font-size:10.0pt;color:black;mso-ansi-language:EN-GB'>> <span
class=GramE>leg (i.e., segment of the Zig <span
class=SpellE>Zag is dynamic,meaning that it
can change.
<span
lang=EN-GB style='font-size:10.0pt;color:black;mso-ansi-language:EN-GB'>>
Therefore, be careful when designing systemtests,
experts, etc. based on
<span
style='font-size:10.0pt;color:black'>> the Zig Zag
indicator."Al Taglavore----------> <span
class=SpellE>From: Owen Davies
<span
style='font-size:10.0pt;color:black'>> <owen@<span
class=SpellE>xxxxxxxxxxxxx>>
<span style='font-size:
10.0pt'>
|