Correlations galore

I'm attaching a spreadsheet with April correlations for all pairs I had access to (31 pairs on Alpari) with corresponding scatterplots.

If you didn't know how to calculate your own correlations, take a look at my spreadsheet and you will be able to do it as well.

I plan to calculate 6 month and 1 year correlations as well, when I'm finished, I'll post the spreadsheets here as well.

Anybody has any other ideas on how we can further explore these besides scatterplots?

edit: I updated the spreadsheet, I included a few extra days from this month into the previous month's calculations by mistake.The new spreadsheet has the error corrected.

thank you, it was a hard job that really needs appreciation, still reading your file and think how can I benefit from them.

actually using this correlation formula would only be useful to find diversification since basically trading highly correlated pairs is just absurd or am I missing something else?

I'm also utilizing the correlation but for trading the lower TFs
usually I would just lined up families of charts of the wideknown correlated pairs

ie. xxx/USD and USD/xxx pairs

and avoid using them during MAJOR fundamental changes

check this pic below and notice the golden candles on the M5 chart

I bet you know what I meant by TRUEly correlated pairs

yes, this is exactly what i plan to do as well.at least in theory, not trading highly inverse correlated pairs should maximize profits (or trading just one pair and not the corresponding inverse correlation )...assuming of course that all the trades on each pair are profitable.

also, again, just in theory, risk could be further reduced if we don't trade positive correlated pairs as well (because if we do, and our analysis for the 1st pair is wrong, we not only loose money on that but on the highly positive correlated 2nd pair we traded as well).

so theoretically, the least risky situation would be to trade only the least correlated pairs, no matter of the correlation type, positive or negative.

the 6 month correlation spreadsheet is ready and i've attached it to the post.

finally, 1 year correlations are also finished.
pay attention to the colored rows and columns in the spreadsheet, i marked those because Alpari has less than 12 months of data available on them so take the correlation coefficients related to those pairs with a grain of salt.

at the end of this month we should have enough data for those as well if i calculated correctly, i will reupdate the spreadsheets then.

I'm uploading an Excel 2007 version of the file as well since I was getting some weird warnings about compatibility...hopefully nothing "funny" happened to the 2003 file format one.

i think i'm going to do a 3 month spreadsheet as well.will post here when i'm done.

yay, 3 month correl done as well spreadsheet attached

This link might save you some work (or maybe not):

http://www.mataf.net/en/forex/tradin...elation/table/

i know about mataf...i don't really trust data that i didn't calculate myself...plus they don't have scatterplots like my spreadsheets and the 1h and daily correlations on that site are something i don't consider very reliable.
i prefer to look at longer time frames for a general idea.

According to more recent research I did, maybe the Spearman Rank Correlation would be more adequate to study than the Pearson one because it can be applied on data samples that do not follow a normal distribution?

The spreadsheets I attached previously all used Pearson correlations.

Has anyone here ever attempted to do this?

Read more here:

http://en.wikipedia.org/wiki/Spearma...on_coefficient

Except it would be a bit more difficult to compute because Excel doesn't seem to have a premade function for it.Maybe I can find a free add-on stat pack... otherwise no problem, will brute-force it by hand.