PureBytes Links
Trading Reference Links
|
Well, I am certain you got the real point there: You do not understand. I am
certain the reason i simple: You do not think, you are too busy defending
tradestation.
The fact that you can get some calculations in EL to show 7 digits is not a
proof that is has 7 digits precision. The fact that there is examples where
it does it worse than single precision calculations is proof it does not
have 7 digits precision. You just do not listen, you just keep on repeating
what you read in the commercials instead of looking at what actually
happens. C does not have 7 digits precision, it has 6, just read the manual
and it will tell you. In spite of that, it does a better job than EL which
has lower precision than C.
I was not comparing two variables in a test, I was delivering an example of
code in Tradestation that did not do its job. I gave to reasons: One being
that calculations in Tradestation does not have same precision as single
precision calculations, the other that the coding done by Omega does not
take this into account. You sometimes use the code I give as an example of
not working code as proof I do not know how to write an indicator. Amasingly
fuzzy logic.
To prove that the calculations are actually delivering 7 digit precision you
deliver some figures where they seem to do, instead of looking at the
examples floating around where they do worse than single precision. That
might have been done by a teacher in physics, but not by anyone knowledgable
in physics. Computing precision is not defined by what it can achieve, it is
defined by the worst thing it achieves.
You are still not teaching me anything, you are just repeatedly ignoring
facts, and repeatedly repeating marketing material.
There is no reason to debate this with you. There is never a good reason to
debate with someone holding his hans over his ears. You will defend
Tradestation in spite of proofs like you have always done.
> -----Original Message-----
> From: pierre.orphelin [mailto:pierre.orphelin@xxxxxxxxxxxxxx]
> Sent: den 17 juli 2001 23:54
> To: Eskimo Omega List
> Subject: RE: Special thread for Pierre Orphelin, the man who
> do not want to see nonsenses again
>
>
> Frankly, I do not understan what and why you are trying to
> demonstrate by dozens of lines of text...
>
> 1) EL has a 7 digit precision, as FLOAT do have in C.
>
> Try this for example:
>
> value1= 555555555;
> value2= value1-444;
> plot1(value2,"");
>
> messagelog(value2:15:15);
>
> Here is the result:
>
> 17/07/2001 23:34 555555136.000000000000000
> 17/07/2001 23:34 555555136.000000000000000
> 17/07/2001 23:34 555555136.000000000000000
> 17/07/2001 23:34 555555136.000000000000000
>
> Means that:
> 17/07/2001 23:34 5 5 5 5 5 5 1 <== 7th digit is
> correct. ===> Garbage
> starts from 8th... 36.000000000000000
>
> FLOAT in C have a 7 digit precision and a +- 1.5 10^+ -34 range.
>
> Your previous test was comparing the garbage part of two
> variable ( dmplus and dmminus flip calculus) ands this is the
> first thing that you do not understand,and the reason of your
> wrong thinking, what you translate by the easy word "bug".
> Yes, the bug exists, but not within TS.
>
> The second one deals with price precision.
> Price have 5 digit precision, no more.
> Any result from the raw data that has more than 5 digit must
> be considered as garbage from the 6th digit. EL precision
> going up to 7 digit, this is enough to have correct results.
>
> Sorry to give you some lessons, but I am a physics teacher.
> And the first thing that we teach to students with
> experimental data is the limit of experimental precision.
> Any measure instrument has it own precision, TS included.
> Suppose that you measure the size ( L x l) of room with a
> ruler, you will get 4 digits, no more for L and l.
>
> Calculate the surface S= L*l.
> If you display more than 4 digits for the surface result,
> you are dead wrong with the extra digits after the 4 th. (I
> simplify here but we should deal with relative errors that
> have additive properties).
>
> This apply to any domain, including technical indicators
> derived from raw
> data:
> Their precision CANNOT be more than the precision of the initial data.
>
> Period.
>
> PO
>
>
> > -----Message d'origine-----
> > De : Bengtsson, Mats [mailto:mats.bengtsson@xxxxxxxx]
> > Envoyé : mardi 17 juillet 2001 19:05
> > À : pierre.orphelin@xxxxxxxxxxxxxx; Eskimo Omega List
> > Objet : RE: Special thread for Pierre Orphelin, the man who do not
> > want to see
> >
> >
> > God, this is taking some time. You do actually not explain
> anything to
> > me, you are just stubbornly confusing people by trying to
> state things
> > against the facts provided by the examples of what happens in real
> > life calculations. (snip)
> >
>
This message contains information that may be privileged or confidential and is the property of the Cap Gemini Ernst & Young Group. It is intended only for the person to whom it is addressed. If you are not the intended recipient, you are not authorized to read, print, retain, copy, disseminate, distribute, or use this message or any part thereof. If you receive this message in error, please notify the sender immediately and delete all copies of this message.
|