|
PureBytes Links
Trading Reference Links
|
<FONT face=Arial color=#0000ff
size=2>Barry,
<FONT face=Arial color=#0000ff
size=2>
Equis
have apparently corrected an error in the calculation of Linear Regression in
Metastock Version 8.0. So if you are running Version 8.0 you may wish to contact
Equis to obtain your Version 8.01 CD-Rom fix disc. Can't explain the anomaly of
ATR but you will appreciate (as Roy pointed-out) that it is possible that
different software programs implement Welles-Wilder's ATR differently given the
unique averaging that Welles-Wilder used.
<FONT face="Kunstler Script" color=#008000
size=5>Regards,
<FONT face="Kunstler Script" color=#008000
size=5>
Gordon
Sutherland
<FONT
face=Tahoma size=2>-----Original Message-----From: Barry Seeger
[mailto:barry.seeger@xxxxxxxxxxx] Sent: Tuesday, 22 April 2003 2:05
p.m.To: equismetastock@xxxxxxxxxxxxxxxSubject: RE:
[EquisMetaStock Group] Inaccurate Indicators
<FONT
face=Arial color=navy size=2><SPAN
>Roy<FONT
face=Arial color=navy size=2><SPAN
>,
<FONT face=Arial color=navy
size=2>Many
thanks for pulling that information together for me. I see now that you were
writing about the difference in calculated values between an Indicator and an
Exploration in MS, and how that can be corrected by forcing the exploration to
use a longer data series.
<FONT face=Arial color=navy
size=2><SPAN
>
<FONT face=Arial color=navy
size=2>My issue
is different from that. It concerns MS calculating a different value for an
indicator that incorporates Linear Regression and
<FONT face=Arial color=navy
size=2><SPAN
>Average<FONT
face=Arial color=navy size=2><SPAN
>
<SPAN
>True<FONT
face=Arial color=navy size=2><SPAN
>
<SPAN
>Range<FONT
face=Arial color=navy size=2><SPAN
> than another program
(Trade Station) does. It has been suggested to me that the other program is
right, and MS gets it just a few cents wrong.
<FONT face=Arial color=navy
size=2><SPAN
>
<FONT face=Arial color=navy
size=2>Any idea
whether that would be the case, and if so, how to correct
it?
<FONT face=Arial color=navy
size=2>Thanks
again,
<FONT face=Arial color=navy
size=2><SPAN
>Barry
<SPAN
>
<SPAN
>
<SPAN
lang=EN-US
>-----Original
Message-----From: Roy Larsen
[mailto:rlarsen@xxxxxxxxxxxxxx] <SPAN
>Sent: <st1:date
Year="2003" Day="21" Month="4"><SPAN lang=EN-US
>Monday,
21 April 2003<SPAN
lang=EN-US
>
<SPAN
lang=EN-US
>1:50
PM<SPAN lang=EN-US
><SPAN
>To:
equismetastock@xxxxxxxxxxxxxxx<SPAN
>Subject: Re: [EquisMetaStock Group]
Inaccurate Indicators
<FONT face="Times New Roman"
size=3>
<FONT face="Courier New"
size=2>Barry<FONT
face="Courier New" size=2><SPAN
><FONT
face="Courier New">Many standard MS indicators use a form of exponential
smoothing as part asan integral
part of the indicator. Exponential moving averages don't
"dropoff" old data in the same
way that Simple moving averages do, so a part of<FONT
face="Courier New">ALL historical data available to the EMA is included in it.
With each newbar some of the new
data is added, this replacing a proportion of the old<FONT
face="Courier New">data, but not ALL. The retained data still has SOME of all
previous data.Addmittedly the
oldest data is just a very small fraction of the current
EMAvalue, but the absence of old
retained data in an exploration that only<FONT
face="Courier New">requires "minimum" periods can cause a significant
variation in value thanthe same
EMA observed on a chart.<FONT
face="Courier New">Suppose your exploration uses "minimum" periods, and the
longest periodindicator isa a 14
period EMA. Now if your charts are set to load 1000
barsthen then a 14 period EMA
will calculate using 1000 bars of data. However<FONT
face="Courier New">the exploration will only be forced to use 14 bars of data.
In such asituation the chart EMA
is much more accurate. Put another way, the<FONT
face="Courier New">exploration EMA could be off by several
percent.The easiest solution
to this problem is to "force" all explorations to
scanat least 5 times the number
of periods of the longest EMA affected<FONT
face="Courier New">indicator. I seldom use less than 400 bars in an
exploration, and never use<FONT
face="Courier New">"minimum" unless I am testing for the very situation
discussed here.What you must
realize is that many indicators include EMA, Wilders, or
someother form of exponential
smoothing as part of the internal calculation.<FONT
face="Courier New">The following post I made at StockCentral some time ago may
also be helpful.See <A
href="">http://www.stockcentral.com.au/forum/machine/Forum32/HTML/000414.html<FONT
face="Courier New">From the number of questions that I have seen posted
recently aboutexploration events
not coinciding with charted events I thought it might
beuseful to discuss further the
major cause of the problem, and to identify<FONT
face="Courier New">many of the MetaStock functions that can contribute to
it.The problem does not lie
with the MetaStock formula language but with the<FONT
face="Courier New">way that individual users set up their
explorations.There seems to
be a general lack of understanding, even among
experiencedusers, that
EXPONENTIAL moving averages require much more data than
the"PERIODS" value of any
affected function will supply. When using the<FONT
face="Courier New">Explorer with the "Minimum Periods" option set we need to
be aware that thenumber of
records explored will not exceed the highest "PERIODS" value
usedin the active exploration
formulas.An EMA displayed on
a chart has access to ALL loaded data, regardless of
the"PERIODS" value of the
indicator. However an exploration will only have<FONT
face="Courier New">access to the amount of data that the user has specified.
The point toremember with an
exponential MA is that new data is added with each bar,
butold data is never dropped out
(as happens with a simple MA). The old EMA<FONT
face="Courier New">data becomes less significant with each new bar but it
continues to be acomponent of the
final EMA value. This is the essential factor that
causesdifferences between chart
and exploration values.Quite
a number of MetaStock functions use internal smoothing, and this
isusually (but not always)
exponential in nature. Some functions allow user<FONT
face="Courier New">control over the smoothing method employed, Mov() being an
obvious example.The following
list cannot be guaranteed as 100% accurate but these are
theMS functions to watch out for
that appear to use, or are in themselves, a<FONT
face="Courier New">form of exponential smoothing.<FONT
face="Courier New">adx()<FONT
face="Courier New">adxr()<FONT
face="Courier New">atr()<FONT
face="Courier New">bbandbot()<FONT
face="Courier New">bbandtop()<FONT
face="Courier New">dema()<FONT
face="Courier New">dx()<FONT
face="Courier New">emv()<FONT
face="Courier New">forecastosc()<FONT
face="Courier New">imi()<FONT
face="Courier New">inertia()<FONT
face="Courier New">macd()<FONT
face="Courier New">mass()<FONT
face="Courier New">mdi()<FONT
face="Courier New">mov()<FONT
face="Courier New">oscp()<FONT
face="Courier New">oscv()<FONT
face="Courier New">pdi()<FONT
face="Courier New">pfe()<FONT
face="Courier New">projosc()<FONT
face="Courier New">qstick()<FONT
face="Courier New">rangeindicator()<FONT
face="Courier New">rmi()<FONT
face="Courier New">rvi()<FONT
face="Courier New">sar()<FONT
face="Courier New">stoch()<FONT
face="Courier New">tema()<FONT
face="Courier New">trix()<FONT
face="Courier New">wilders()<FONT
face="Courier New">When using any of these functions in an exploration the
accuracy of thereported results
will require the number of records explored to be set<FONT
face="Courier New">appropriately. As a rule of thumb I would suggest
multiplying the highest"PERIODS"
value by a factor of 5.<FONT
face="Courier New">Another trick that should force the exploration to include
the requiredminimum number of
records without actually specifying that number is to
addthe following code to the
filter and adjust "PERIODS" to a suitable value.<FONT
face="Courier New">" AND
Mov(C,PERIODS,S)>0"<FONT
face="Courier New" size=2><SPAN
>Roy<FONT
face="Courier New" size=2><SPAN
><FONT
face="Courier New">> I recall that Roy Larsen has responded on this
listserv to queries about> why
indicators in MetaStock can give inaccurate results. From memory
it> occurs when insufficient
data is loaded.<FONT
face="Courier New">>> I
didn't take much notice of the discussion at the time, but I have
now> experienced that
problem.<FONT
face="Courier New">>> I
can't see a document covering the issue with this group's files, yet
I> recall there was such a
document. Can Roy or anyone direct me to it<FONT
face="Courier New">> please?<FONT
face="Courier New">> Thanks,<FONT
face="Courier New">> Barry<FONT
face="Courier New">><FONT
face="Courier New">><FONT
face="Courier New">><BR
><BR
><FONT
face="Courier New" size=2>To unsubscribe from
this group, send an email to:<FONT face="Courier New"
size=2><FONT
face="Courier New">equismetastock-unsubscribe@xxxxxxxxxxxxxxx<FONT
face="Courier New" size=2>Your use of Yahoo!
Groups is subject to the Yahoo!
Terms of Service.
To
unsubscribe from this group, send an email
to:equismetastock-unsubscribe@xxxxxxxxxxxxxxxYour
use of Yahoo! Groups is subject to the <A
href="">Yahoo! Terms of Service.
Yahoo! Groups Sponsor
To unsubscribe from this group, send an email to:
equismetastock-unsubscribe@xxxxxxxxxxxxxxx
Your use of Yahoo! Groups is subject to the Yahoo! Terms of Service.
|