|
PureBytes Links
Trading Reference Links
|
Ron:
I have enjoyed your posts for some time and I really like this concept of
data
comparison. You talk about a 50% correlation which brought up a question
of
just how detailed do you get in the correlation of data. How many
elements do
you use in the comparison and are these elements compared against every
bar in
the sequence? If so, it seems like a formidable task. Is there a list of
comparisons that you could share in order to get an idea of how to do
this type of analysis? Do you weight all the comparisons equally?
Thanks for any help and keep up the good work.
Lynn Green
The correlation study is very simple technique. It is done in Excel using
the "CORREL" function. I set up a spread sheet with the date, open, high,
low, close data for the OEX. I used five columns of correlation formulas
3 days, 5 days, 8 days, 13 days and 21 days. to get the 3 day correlation
you drag in the open, high, low and close for the three days are
interested in for the first part of the correlation function, then you
drag the same data series for the second part of the correlation
function. Anchor the first data series ie =correl(A$3$:D$5$,A3:D3).
Repeat the same for the other time frames. Drag and fill this formula
into the rest of the spreadsheet. You can jazz this up with some
database filter retrieval function or you can simply scan the list for
rows where the numbers are all over 50%. You would have a better than 50%
correlation with a 3 day, 5 day, 8 day 13 day and 21 day, open, high,
low and close pattern. When this is all done and you have your pattern
matches apply some indicators that you like to the different data matches
to try and confirm the similarities in the developing pattern. This is as
close to having tomorrows Wall Street Journal as you will get (legally !)
Happy New Year!
Ron McEwan
|