|
PureBytes Links
Trading Reference Links
|
> -----Message d'origine-----
> De : MikeSuesserott [mailto:MikeSuesserott@xxxxxxxxxxx]
> Envoye : dimanche 29 juillet 2001 20:35
> A : pierre.orphelin@xxxxxxxxxxxxxx; omega-list@xxxxxxxxxx
> Objet : AW: TradeStation Precision - Summary
>
>
>
> I am sure you are aware of the difference between repetition and
> iteration. Your test just *repeats* the same brief iteration with the same
numbers
> 35000 times. Not much enlightenment is to be expected from such a
> test, and you yourself correctly named it "debile". <g>
>
OK, OK.
I want to sell a Safir-X copy to you that will be compatible with
Mathematica( it's not a joke, we can do it).
So here is a test that uses variable data and shows the iteration effect
with different numbers.
The code below will produce a 1000 bars period sine curve, where we apply
the xaverage fuction.
We check the xaverage value every 1000 bars, that is of course expected to
be the roughly same ( it cannot be exactly the same because of the EMA
structure as you pointed out)
value1= cosine( 360/1000*barnumber);
value2=xaverage(value1,10);
if mod(barnumber,1000)=0 then messagelog(barnumber:5:0," ",value2:1:8);
plot1(value1);
The results are below:
Date Content Analysis Technique
30/07/2001 00:17 30000 0.99908400 Indicator: test debile
30/07/2001 00:17 29000 0.99908500 Indicator: test debile
30/07/2001 00:17 28000 0.99908600 Indicator: test debile
30/07/2001 00:17 27000 0.99908700 Indicator: test debile
30/07/2001 00:17 26000 0.99908800 Indicator: test debile
30/07/2001 00:17 25000 0.99908900 Indicator: test debile
30/07/2001 00:17 24000 0.99909000 Indicator: test debile
30/07/2001 00:17 23000 0.99909100 Indicator: test debile
30/07/2001 00:17 22000 0.99909200 Indicator: test debile
30/07/2001 00:17 21000 0.99909300 Indicator: test debile
30/07/2001 00:17 20000 0.99909400 Indicator: test debile
30/07/2001 00:17 19000 0.99909500 Indicator: test debile
30/07/2001 00:17 18000 0.99909500 Indicator: test debile
30/07/2001 00:17 17000 0.99909600 Indicator: test debile
30/07/2001 00:17 16000 0.99909700 Indicator: test debile
30/07/2001 00:17 15000 0.99909800 Indicator: test debile
30/07/2001 00:17 14000 0.99909900 Indicator: test debile
30/07/2001 00:17 13000 0.99910000 Indicator: test debile
30/07/2001 00:17 12000 0.99910100 Indicator: test debile
30/07/2001 00:17 11000 0.99910200 Indicator: test debile
30/07/2001 00:17 10000 0.99910300 Indicator: test debile
30/07/2001 00:17 9000 0.99910400 Indicator: test debile
30/07/2001 00:17 8000 0.99910500 Indicator: test debile
30/07/2001 00:17 7000 0.99910600 Indicator: test debile
30/07/2001 00:17 6000 0.99910700 Indicator: test debile
30/07/2001 00:17 5000 0.99910800 Indicator: test debile
30/07/2001 00:17 4000 0.99910900 Indicator: test debile
30/07/2001 00:17 3000 0.99911000 Indicator: test debile
30/07/2001 00:17 2000 0.99911100 Indicator: test debile
30/07/2001 00:17 1000 0.99911200 Indicator: test debile
After 30,000 bars of iterations, the error compared to iteration 1000 is:
0.99911200-0.99908400 = 0.000028.10^-5, so it affects the 5th digit.
Now, more research has to be conducted to see what is the part of the EMA
calculus itself in the error, and what is the part of the TS float
precision.
In any manner, this difference is not so big as you may expect (3*10^-5)
after 30,000 bars, still acceptable with precision prices in most cases, but
could be better...
The order form is on the web site below
Sincerely,
Pierre Orphelin
www.sirtrade.com
TradeStation Technologies representative in France
|