|
PureBytes Links
Trading Reference Links
|
> -----Message d'origine-----
> De : MikeSuesserott [mailto:MikeSuesserott@xxxxxxxxxxx]
> Envoye : lundi 30 juillet 2001 12:29
> A : pierre.orphelin@xxxxxxxxxxxxxx; omega-list@xxxxxxxxxx
> Objet : AW: TradeStation Precision - Summary
>
>
> Hello Pierre,
>
> thanks for your messages, and for trying to find a good test.
>
> I believe the main problem might be that it may be difficult if not
> impossible to ferret out these kinds of instabilities by means of a test
> done within TS alone. Let me think of something that we can do in
> parallel - you can do in EL and I will run it in Mathematica. Probably
nothing
> trade-specific (so that you might be satisfied it does not depend on any
> "buggy libraries" of mine), but rather something along the lines of your
> cosine example. I will disclose my piece of code here on the
> list, and then we can discuss how to proceed.
>
> I shall probably not have time for this during the week, but would post my
> code and suggestions over the next weekend. D'accord?
I may suggest a simple test:
Consider a EMA crossover average system tested with 30,000 bars on an
intraday continuous contract and giving a lot of trades, (buy on close, sell
on close only to avoid other discussion).
The system will be written in order to only consider for final decision the
5 digits (all digits available rounded to 5 digits) produced by the
difference of the EMAs, 5 digits in TS, 5 digits in Matlab, as we know that
the garbage is for sure after 5 digit, and that we must stay compliant with
the price precision.
Compare the performance summary for both and evaluate the difference if
any.
The order of magnitude of the difference in the performance summary fields
will tell for sure if TS is worth to be recoded using doubles or not when
using recursive calculus that propagate the error like it is the case with
the EMA.
Doing so, the difference will only deal with the need or not of extra
precision.
This is not no much time consuming and I do not want to spend too much time
on this issue.
IMHO, a slight difference could appear in the performance summary results,
but not mission critical.
>
> Michael Suesserott
>
> PS. I think I should apologize to you for that "epater le
> bourgeois" part of my message. I never did believe that you wanted to
mislead the
> public in any way, and I know you are an honest guy and a fine person.
Only at
> times a bit stubborn... <g>
>
Thanks a lot.
I'am always stubborn, every time, every day and in every activity that I
consider of interest to me.
For example, a product like Safir-X and the next Safir-Pro and 4 versions
to come would never have existed if I was not.
Sincerely,
Pierre Orphelin
www.sirtrade.com
TradeStation Technologies representative in France
|