RE: From Spread's to indexes

["c" <camacazi@xxxxxxxxxxx>]

Thanks to all with the responces to my index enquires

:O)


Cheers
Cameron













-----Original Message-----
From: drwar [mailto:drwar@xxxxxxxxxxx]
Sent: Monday, March 29, 2004 2:08 PM
To: camacazi@xxxxxxxxxxx; drwar; Gary Fritz; omega-list@xxxxxxxxxx
Subject: RE: From Spread's to indexes


Cameron
You need to normalize the data for each variable first. Then you can combine
them in the proportions you deem appropriate. That can ultimately result in
a simple weigthing factor multiplied times each data item which takes into
account both the normalization and the proportion. The problem you run into
without normalization is the following.

It item ones average value is 2 and item twos average value is 500 a 1
percent change in item 2 will swamp any changes in item 1. Thats why if we
normalize the data to place them all on the same scale and then determine an
appropriate proportion to combine them you will find it a little easier to
work with. Once you know the scaling factor to normalize and the proportion
you want you can combine it into one operation which is just a weight and
the equation will take the form of

Myindex = Ax + By + Cz

where A,B,C = final weighting factors and x,y,z are the data items.

Jerry


-----Original Message-----
From: c [mailto:camacazi@xxxxxxxxxxx]
Sent: Sunday, March 28, 2004 10:23 PM
To: drwar; Gary Fritz; omega-list@xxxxxxxxxx
Subject: RE: From Spread's to indexes


Jerry

if i add an addition 10000 'points' top the myindex variable at the first
bar.... it looks like an index when i plot it  :O)
maybe i need a bigger formula ..one that looks more complicated  :O/
...ummmmmmmmmm maybe ?

cheers
Cameron

-----Original Message-----
From: drwar [mailto:drwar@xxxxxxxxxxx]
Sent: Monday, March 29, 2004 1:14 PM
To: camacazi@xxxxxxxxxxx; Gary Fritz; omega-list@xxxxxxxxxx
Subject: RE: From Spread's to indexes


The code you show looks like a calculation for a weighted 1 bar ROC. I did
not gather that was what you were looking for in your original post.

Jerry

-----Original Message-----
From: c [mailto:camacazi@xxxxxxxxxxx]
Sent: Sunday, March 28, 2004 10:05 PM
To: Gary Fritz; omega-list@xxxxxxxxxx
Subject: RE: From Spread's to indexes


Thanks Gary

Heres basically what i have done for now.....


 d1percent=((close[1] of data1 - close of data1) / close of data1 )* 40/100;
 d2percent=((close[1] of data2 - close of data2) / close of data2 )* 30/100;
 d3percent=((close[1] of data3 - close of data3) / close of data3 )* 30/100;

and then

myindex=myindex+d1percent+d2percent+d3percent;


Works a treat.... any other ideas for improvement just let me know  :O)

Cheers
Cameron



-----Original Message-----
From: Gary Fritz [mailto:fritz@xxxxxxxx]
Sent: Sunday, March 28, 2004 1:58 PM
To: omega-list@xxxxxxxxxx
Subject: Re: From Spread's to indexes


> i want to create my own index.... and i want to weight the individual
> components eg
> heating oil at ummmmmmmm say 40%
> crude oil  at 30%
 DAX at 30%

Set up HO in Data1, CL in Data2, DAX in Data3.
Now you could simply combine them by plotting (however you like):
 0.40*Close of Data1 + 0.30*Close of Data2 + 0.30*Close of Data3

But that doesn't take their widely varying prices into account,
so e.g. CL (at about $35) will have almost 40x larger impact than
HO (at about $0.90).  So you might want to scale them somehow,
maybe something like:

{ Get recent average value }
HO = Close of Data1;
CL = Close of Data2;
DAX = Close of Data3;
HOavg = average(HO, 100);
CLavg = average(CL, 100);
DAXavg = average(DAX, 100);
Index = 0.40*(HO/HOavg) + 0.30*(CL/CLavg) + 0.40*(DAX/DAXavg);

...or whatever calculation suits your fancy.
Gary


---
---


---
---


---
---