|
PureBytes Links
Trading Reference Links
|
Guy,
Too much Stroth's beer at an early age.
Steve Karnish
Cedar Creek Trading
http://www.cedarcreektrading.com
----- Original Message -----
From: Guy Tann <grt@xxxxxxxxxxxx>
To: Metastock User Group <metastock-list@xxxxxxxxxxxxx>
Sent: Thursday, August 24, 2000 9:11 PM
Subject: Precision and MS calculations
> List,
>
> Well I gave up and went back to square one. I couldn't figure out how a
> variable could calculate correctly from 12/11/99 to 8/9/00 and then fail
the
> next day and then work correctly again. First I wrote 7 Clipper programs
> which I used to print out each variable. I checked my MS calculations for
> each variable against these printouts. The first 8 variables calculated
> correctly.
>
> The ninth variable didn't! What was weird was that every day calculated
> correctly until the 2nd day from the end. I double-checked the inputs and
> they were in complete agreement (I exported my MS data into Clipper to
> insure both systems that were using the same data). I then went back and
> double-checked each value that was part of this calculation. One of my
> variables was off slightly for one of these days. I went back and recoded
> my MS code and broke the calculation into 6 steps to equal the code in the
> Clipper program. So now I have identical code in both modules and you
would
> think that the results would be identical. Wrong, buckwheat!
>
> For the life of me, I'm at a loss. It's probably something to do with MS
> and precision, but in this case, following rounding, I use Int() to
capture
> only the whole number and drop any fractions. If anyone has a suggestion,
> I'm open to it. Our numbers are too sensitive to not be able to trust the
> results of a simple calculation. If this was a complex calculation, I
might
> possibly understand a variance. In this case the correct answer is -3
(and
> yes, I did calculate it manually to make sure) while MS came up with -2.
> I think I'm going to take a day off and clear my head. This is why I get
so
> frustrated trying to use MS. How could a program calculate correctly for
> 99.994% of the time and then blow it on that one day?
> Tomorrow, I'm planning on printing out the intermediate results from both
> Clipper and MS to see if I can see where this thing is going wrong. I'm
too
> tired tonight.
>
> I need a break. Where did I put that beer?
>
> Guy
>
> Never be afraid to try something new. Remember, amateurs built the ark,
> professionals built the Titanic.
>
|