|
PureBytes Links
Trading Reference Links
|
> -----Message d'origine-----
> De : Bengtsson, Mats [mailto:mats.bengtsson@xxxxxxxx]
> Envoyé : dimanche 29 juillet 2001 23:02
> À : pierre.orphelin@xxxxxxxxxxxxxx; Omega-list
> Objet : RE: TradeStation Precision - Summary
>
>
> > You should try to buy it and see if we accept your
> > improvements, if any! Usually we take in account all the
> > hints from our customers and add them to the next version if
> > we can prove us that it is of benefit to them. This is very simple.
>
> Sounds like true Omega thinking. We as users will not get division by zero
> checks or working DMIs, since we only believe we need it, but we have not
> proven it to Omega (like Pierre said about the DMI "how much did
> you really> lose"). Same for your software, all that it takes is to prove
to you
> something that you might not want to believe. That sounds like a
> good reason not to believe in updates from you driven by users, you are
not thinking
> openly, rather you are fighting ta maintain your points, which may be
> different points in different letters on the same topic.
No,nononono....
Buy it,try it for free, then you will see.
>
> In some mails you write about single precision float being more than
> adequate, all miscalculations shown to you are irrelevant. In
> other letters
> you claim that the Bob Fulks solution with recalculating for example
> averages every 100th bar or so would cause big jumps in the values when
> recalculations occur... :-)
Yes, IF the 100 th reset comes after the sum is more than single precision
(averaging big number like volumes or price* volume), the jump could occur.
The arbitrary reset without control of when resetting MAY produce this
effect and I proposed an alternate solution ( that you forgot as well)where
I checked the reset condition.
But this was an issue with the average function and the improvement proposed
by Bob, not the Xaverage.
Do not mixup all of what I say, or try to avoid to truncate.
>
> Then you are back again in other letters, saying that Omegas version of
> single precision is fully adequate for a physician, the low precision in
> data means that it is irrelevant if the calculation process
> introduces extra errors during the calculations due to their design.
Misunderstanding.
TS precision is enough for calculation with 5-6 digits prices in technical
analysis, as long as the errors are beyond this precision, regardless the
job of the user
>
> Ah well, just trust that you are open, only thing needed is just
> to prove to you that your users really need what they think they want.
>
I just try to help.
> Easy thing... :-)
>
Not so easy...
Sincerely,
Pierre Orphelin
www.sirtrade.com
TradeStation Technologies representative in France
|