Information For Beginners

They don't have complete data to check against - another limitation of using a website.

Again, near the mark even if off by a month.

if you draw up truth tables (or trade tables?) for all three data sets they will suggest trading one pair from them at any one time. so you can still follow all tightly correlated currencies such as usd majors if you want to.

i think that on the usd's then perhaps you can trade usdjpy and usdchf fairly independently.

this would be pretty good but the eurusd has a high correlation with gbpjpy so they couldn't be traded together.

just to show really how highly correlated jpy pairs can be - it would be just as bad to follow jpy majors like usd majors i think.

this is one i would think about, after adding back eurusd (not shown):

true < 50% (50/50 without error correction)

true < 60% (including 10% random error correction) - what the pundits recommend.

see how the numbers of t's increases giving you more choice. i think this table is the best i've come up with so far. can you beat it..?

is the whole thing academic?

well, yes and no - depending on how you do it.

take no notice of correlation then you might commit financial suicide if you are over exposed to a particular type of market - e.g usd majors around nfp time for example.

go to extremes with up to the minute correlations then you make the whole thing too complex to manage and decide from.

pick a middle way - say do the whole exercise every 6 months, to a year or 5 years. Beyond 5-10 years I don't think there's much point since the older data might become to irrelevant as markets change by nature. What was true back then might be irrelevant now.

Do your own studies, come up with your own solution. You can't just simply ignore the problem if you trade more than one market (or pair).

It's a continual problem traders have to manage somehow.

I got the historical data from strategybuilderfx as far back as the charts would go and exported the daily data to csv format.

Attach the CSV tables to the Access 2003 database. You'll have to delete the existing links then reattach the tables. Make sure you select the comma option and make sure 7 fields are picked up. Don't rename anything, just accept all the defaults.

The excel spreadsheet is what you're after. Each tab has data imported from the queries in the database. They should match up by query number. Go to tools->import data to see how it's done.

The tab at the end, shows how to use the correl function to work out the correlations. Basically find the date (field 1) and then the row number and use the row numbers as a date range for correlation comparisions.

The second zip contains the mataf reports for verification.

For those with a programming background check it all out - you have the same data that I used. And now you can do it all yourself without relying on questionable tables from so called "official" sources.

Best of luck!

I have been doing research on this for a long time now.

Though the correlations are considered accurate, there have been anamolies.
I think the attached chart should explain it self

Correlation between eur/usd and gold

Hey Iso,

Thanks for sharing. Now I have more functionalities to learn and play with Excel and Access.

Thanks agian.

East

this pairing is next on my list, i think the chart is more of a visual correlation thing and it looks like quite a high correlation there.

my guess would be 70-90% before i crunch the numbers...

Have fun dude!

Both samples were written using Office 2003, but they might work with 2000 as well.

For non development people then the spreadsheet should be all you need really. The data is in the first 7 tabs and the last one shows you how you use the correl function. Most of the major groups have been covered so there should be no need to bring in new data from the Access database, or from the CSV files.

Are you going to add intermarket Correlations too..

If by intermarket you mean SP500 and Gold then for sure!!!

Here we go, looks like EURUSD goes with just about anything...

i think i might settle for this portfolio from now on, so that means booting out GBPCHF for GBPJPY, and also EURJPY for EURCHF.

after playing around with the data, i can't find anything better really.

so the table reads with true=60%.

I think the dow jones utility average should be substituted for the S&P 500 as it reacts to inflation faster than the S&P 500